CN103370004B - Cardiac decompensation detection using multiple sensors - Google Patents

Abstract

Physiological data, such as thoracic impedance data, may be obtained during a first time window to establish a baseline, or may be used to form one or more data clusters. Additional physiological data, such as thoracic impedance test data obtained during a later time window, may be obtained and compared to a baseline or data cluster to determine an indication of worsening heart failure. In one embodiment, the quantitative attributes of one or more data clusters may be monitored and used to provide an indication of worsening heart failure. A posture discrimination metric may be obtained, such as using physiological data obtained during a first time window. Additional physiological data, such as may be obtained during a second time window, may be compared to the posture discrimination metric to provide a posture state of the patient.

Description

Cardiac decompensation detection using multiple sensors

priority claim

This application claims priority benefit under 35 U.S.C. §119(e) to U.S. Provisional Patent Application Serial No. 61/423,127, filed December 15, 2010, by Thakur et al., entitled "Cardiac Decompensation Detection Using Multiple Sensors," It is incorporated herein by reference.

Background technique

Cardiac rhythm management devices may include implantable or other ambulatory devices such as pacemakers, cardioverter-defibrillators, cardiac resynchronization therapy (CRT) devices, or may monitor one or more physiological parameters, or provide pacing , defibrillation, or cardiac resynchronization therapy, or a device that can monitor one or more physiological parameters and provide therapy. In one embodiment, such a device may be configured for use with implantable or external electrodes, such as to detect or treat cardiac, respiratory or other conditions. Information such as obtained using the electrodes can be used to provide a diagnosis or predict an impending disease state, or to initiate or adjust therapy.

Early detection of physiological conditions that can be indicative of heart failure ("HF", sometimes called congestive heart failure, "CHF"), such as before a patient experiences cardiac decompensation associated with such heart failure, can help indicate possible Treatment to prevent the occurrence of cardiac decompensation. Zhao et al., in U.S. Patent Publication No. 2010/0004712, entitled "SYSTEMS AND METHODS FOR USE BY AN IMPLANTABLE MEDICAL DEVICE FOR DETECTING HEART FAILUREBASED ON THE INDEPENDENT INFORMATION CONTENT OF IMMITANCE VECTORS)" mentions the detection of heart failure based on the independent information content of multiple impedance vectors, and the volume control function based on the independent information content. (See Zhao et al., at paragraph [0008].)

Accurate and efficient postural detection can help provide important information to a clinician or cardiac rhythm management device, for example, to ensure accurate interpretation of one or more physiological parameters, or to determine therapy. Maile et al., in U.S. Patent No. 7,559,901 entitled "DETERMINING A PATIENT'S POSTURE FROM MECHANICAL VIBRATIONSOF THE HEART", mention determining the patient's position by monitoring heart sounds. (See Maile et al., at "Abstract.")

overview

Various electrical or mechanical functions of the heart can provide a variety of physiological parameters that can indicate the onset of conditions such as heart failure, cardiac arrhythmias (fibrillation, tachycardia, bradycardia), ischemia, and the like. These physiological parameters may include, for example, heart sounds (eg, S3 amplitude), DC impedance near the lungs, heart rate, respiration rate, weight, or intracardiac pressure. Additional examples of physiological parameters may include, but are not limited to, hormone levels, blood counts, nerve activity, muscle activity, or any other physiological parameter. At least some of these parameters may be indicative of an onset or change in a condition, which may be used to provide an alert that therapy (or therapy adjustment) is required, such as defibrillation, pacing changes, and the like. However, it is difficult to determine whether an event is starting when only a few measurements for these parameters indicate the onset of a disorder.

Among other things, this document describes systems, methods, machine-readable media, or other techniques that may involve obtaining physiological data, sorting physiological data into one or more data clusters or establishing a baseline, obtaining additional test data, and comparing the test data to Data clustering or baseline comparison to identify indications of cardiac decompensation.

The technique may involve obtaining physiological data over a first time window to establish a baseline, which may include two or more discrete groups, where one or more groups may correspond to patient positions. Additional physiological data can be obtained and compared to the group to provide the patient's postural status.

The techniques described and illustrated herein can be directed towards diagnosing a patient at risk for cardiac decompensation prior to heart failure. Likewise, some posture detection techniques are proposed. The techniques described and illustrated herein may likewise or alternatively be used to determine or differentiate among one or more of pulmonary edema, pleural edema, or peripheral edema.

The present technique may provide one or more indicators of cardiac decompensation, which may provide improved specificity, such as relative to other ways of predicting cardiac decompensation using only raw impedance data. Impedance data can be modulated in response to normal neurohormonal or circadian rhythm changes in the patient. Therefore, it may be difficult to discern trends in impedance data. Also, systemic changes in patient fluid levels can affect the overall measured impedance vector at once, which can confound decompensation detection based solely on the raw thoracic impedance data used to determine the effusion. Rather, the present technique may include, among other things, comparing at least two impedance vectors, which may include at least one body position change, such as during one or more time windows. In such an approach, effects due to changes in systemic fluid levels can be minimized. This could help provide a better indication of an increased risk of heart failure decompensation.

The inventors have identified that, among other things, the problem to be solved may include providing more sensitive or specific advance notification of a patient's impending risk of cardiac decompensation. In embodiments, the subject matter can provide a solution to this problem, for example, by obtaining physiological data, forming a function of the first physiological data and the second physiological data, identifying one or more trends in the physiological data, and using the The information provides advance notice of the patient's risk of cardiac decompensation.

The inventors have identified that, among other things, another problem to be solved may include providing the patient's postural status, for example using physiological data readily available to an implantable cardiac rhythm management device—e.g., without the need for a dedicated 3-axis Accelerometers, tilt switches, or other sensors to detect patient orientation or position. In embodiments, the subject matter can provide a solution to this problem, for example, by using one or more chest impedance measurements to establish a postural discriminant comparison metric, obtaining chest impedance test data, and by comparing the chest impedance test data with the postural discriminant comparison metric Postural status is provided for comparison.

This summary is intended to provide an overview of the subject matter of this patent application. It is not intended to provide an exclusive or exhaustive explanation of the invention. The Detailed Description is included to provide further information about this patent application.

The present invention relates to the following embodiments:

1. A medical device comprising:

a processor, the processor comprising:

a first data input configured to receive first physiological data from a first physiological sensor, the first physiological data corresponding to a plurality of instances during a first time window; and

a second data input configured to receive second physiological data from a second physiological sensor, the second physiological data corresponding to a plurality of instances during the same first time window; and

A processor-readable medium comprising instructions that, when executed by the processor, configure the medical device to:

forming a function of said first physiological data on said second physiological data;

forming at least two data clusters using the function;

determining a quantitative attribute of at least one of the data clusters;

The quantitative attributes are used to provide an index of heart failure decompensation.

2. The medical device of embodiment 1, wherein the first data input is configured to receive first physiological data comprising a first chest impedance obtained using a first electrode configuration defining a first chest impedance vector data; and

Wherein the second data input is configured to receive second physiological data comprising second chest impedance data obtained using a different second electrode configuration defining a different second chest impedance vector.

3. The medical device of any one of embodiments 1 or 2, wherein the first data input is configured to receive first physiological data comprising using a first first electrode, and at least a different second electrode to obtain the first chest impedance data.

4. The medical device of any one of embodiments 1-3, wherein at least one of the first data input or the second data input is configured to be coupled to an accelerometer to obtain the first physiological data or the The second physiological data.

5. The medical device of any one of embodiments 1-4, wherein the processor-readable medium includes instructions that, when executed by the processor, configure the medical device to use body position information to form the At least two data clusters.

6. The medical device of any one of embodiments 1-5, wherein the processor-readable medium includes instructions that, when executed by a processor, configure the medical device to use body position information to determine the second The first and second physiological data are used to form the function.

7. The medical device of embodiment 6, wherein the first data input is configured to receive first physiological data from a first physiological sensor, the first physiological data corresponding to a plurality of instances during a first time window, comprising At least one patient position change.

8. The medical device of any one of embodiments 1-7, wherein the processor-readable medium includes instructions that, when executed by the processor, configure the medical device to use time-of-day information to determine the The first and second physiological data are used to form the function.

9. The medical device of embodiment 8, wherein the first data input is configured to receive first physiological data from a first physiological sensor, the first physiological data corresponding to a plurality of instances during a first time window, The first time window includes a time interval in which a change in patient position is expected to occur.

10. The medical device of any one of embodiments 1-9, wherein the processor-readable medium includes instructions that, when executed by a processor, configure the medical device to determine at least A quantitative attribute comprising at least one of the following:

expansion of parts of said function;

the scope of the portion of the function;

an area defined at least in part by a portion of said function;

the location or centroid of the portion of the function;

the distance between the positions or centroids of at least two different parts of said function; or

A different second function is formed using at least two different parts of the function.

11. The medical device of embodiment 10, wherein the processor-readable medium includes instructions that, when executed by a processor, configure the medical device to compare the baseline distance with the at least two different parts of the function Distance comparison between locations or centroids.

12. The medical device of any one of embodiments 1-11, wherein the processor includes a third data input configured to receive additional physiological data from at least a third physiological sensor, the additional physiological the data correspond to multiple instances during the same first time window; and

wherein the processor-readable medium includes instructions that, when executed by a processor, configure the medical device to form a multidimensional function using the first physiological data, the second physiological data, and the additional physiological data .

13. The medical device of embodiment 12, wherein said processor-readable medium comprises instructions that, when executed by a processor, configure said medical device to determine a quantitative attribute of at least one of said data clusters, comprising The volume is determined using parts of the multidimensional function.

14. The medical device of any one of embodiments 1-13, wherein the first data input is configured to receive third physiological data from the first physiological sensor, the third physiological data corresponding to multiple instances during the window; and

wherein the second data input is configured to receive fourth physiological data from the second physiological sensor, the fourth physiological data corresponding to a plurality of instances during the same second time window; and

wherein the processor-readable medium includes instructions that, when executed by a processor, configure the medical device to

forming a second function of said third physiological data on said fourth physiological data;

forming at least two additional data clusters using the second function;

determining a test quantitative attribute of at least one of the additional data clusters; and

The quantitative properties of the test are used to provide an index of heart failure decompensation.

15. The medical device of embodiment 14, wherein the processor-readable medium includes instructions that, when executed by a processor, configure the medical device to trend the quantitative attribute and the test quantitative attribute to Provides indicators of decompensation in heart failure.

16. The medical device of any one of embodiments 1-15, wherein the processor-readable medium includes instructions that, when executed by a processor, configure the medical device to use time-of-day information to form the at least two data clusters.

17. A system comprising:

Implantable medical devices, including:

a first physiological sensor configured to receive first physiological data corresponding to a plurality of instances during a first time window;

a second physiological sensor configured to receive second physiological data corresponding to multiple instances during the same first time window; and

a processor circuit configured to:

receiving said first and second physiological data;

forming a function of said first physiological data on said second physiological data;

forming at least two distinct data clusters using the function;

determining a quantitative attribute of at least one of the data clusters; and

An index of heart failure decompensation is provided using the quantitative attributes.

18. The system of embodiment 17, comprising an impedance measurement circuit configured to use the first physiological sensor to receive at least a first impedance signal and to use the second physiological sensor to receive at least a first impedance signal. Two impedance signals, wherein the first physiological sensor includes a first electrode and the second physiological sensor includes a second, different electrode.

19. The system of any one of embodiments 17 or 18, the system comprising a storage circuit configured to store a plurality of quantitative attributes.

20. A medical device comprising:

a processor, the processor comprising:

a first data input configured to receive first chest impedance data from a first chest impedance vector, the first chest impedance data corresponding to a plurality of instances during a first time window;

a second data input configured to receive second chest impedance data from a second chest impedance vector, the second chest impedance data corresponding to a plurality of instances during the same first time window;

a third data input configured to receive test chest impedance data from at least two chest impedance vectors; and

A processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to:

forming a first function of said first chest impedance data on said second chest impedance data;

forming a first pair of data clusters using the first function;

determining a first quantitative attribute using the first pair of data clusters;

forming a second function using the test chest impedance data;

forming a second pair of data clusters using the second function;

determining a second quantitative attribute using the second pair of data clusters; and

A comparison of the first and second quantitative attributes is used to provide an index of heart failure decompensation.

Description of drawings

In the drawings, which are not necessarily to scale, like numerals may depict like components in the different views. The same number with a different letter suffix may represent different instances of the same component. By way of example, and not by way of limitation, the drawings generally illustrate various embodiments discussed herein.

Figure 1 generally illustrates an embodiment of a portion of a system, which may include an implantable or ambulatory medical device, and one or more implantable leads or other electrodes, such as may be configured to at least partially communicate with cardiac tissue Associate targeting.

Figure 2A generally illustrates an example of a graphical representation of chest impedance data that may be monitored or recorded over a period of days.

Figure 2B generally illustrates an example of a graphical representation of chest impedance data that may be monitored or recorded over a period of days.

Figure 3 generally illustrates a graphical representation of chest impedance data and body position data that may be obtained over a period of several days.

4A generally illustrates a graph of a function that may plot first chest impedance data from a first impedance vector against second chest impedance data from a second impedance vector.

FIG. 4B generally illustrates a graph of a function that may include one or several quantitative properties of the function, such as centroid location, spread, or extent.

FIG. 4C generally illustrates a diagram of a numerical integration method that can be used to calculate the area of one or more data clusters.

FIG. 4D generally illustrates a diagram of a method that may be used to calculate the area of one or more data clusters, for example, by finding the extent of the data and drawing rectangles.

Figure 5 generally illustrates a second plot of chest impedance data and body position data over a period of days.

FIG. 6 illustrates generally a graph of a second function including several quantitative properties of the function.

Figure 7 generally illustrates an embodiment that may include the use of data cluster attributes to provide heart failure decompensation indicators.

FIG. 8 generally illustrates an embodiment, which may include determining first and second attributes using respective first and second functions of different physiological data, obtaining first and second physiological data during a first time window, converting the The first and second attributes are trended and the trends are used to provide an index of heart failure decompensation.

9 generally illustrates an embodiment, which may include obtaining a postural discriminative metric, obtaining thoracic impedance test data, comparing the thoracic impedance test data to the postural discriminative metric, and providing a postural status.

FIG. 10 generally illustrates an embodiment that may include forming a function of first physiological data versus second physiological data and providing a postural state using a comparison of the physiological test data and the function.

11 generally illustrates an embodiment that may include determining a first quantitative attribute using a first function of first physiological data on second physiological data and determining a second quantitative attribute using a second function of third physiological data on fourth physiological data. attribute, and the postural state is determined using a comparison of the first and second quantitative attributes.

Detailed description of the invention

Physiological data such as chest impedance data may be obtained during the first time window to establish a baseline, or may be used to form one or more data clusters. Additional physiological data, such as chest impedance test data obtained during a later time window, can be obtained and compared to a baseline or data cluster to determine indications of worsening heart failure. In one embodiment, quantitative properties of one or more data clusters may be monitored and used to provide an indication of worsening heart failure.

FIG. 1 generally illustrates an embodiment of a system 100 , which may include an implantable or other ambulatory medical device, such as a cardiac rhythm management (CRM) device 102 . In an embodiment, the CRM device 102 may include an implantable electronics unit 105 . In an embodiment, electronics unit 105 may be electrically and physically connected to implantable lead system 110 .

Portions of implantable lead system 110 may be inserted into the patient's chest, including into the patient's heart 107 . Implantable lead system 110 may include one or more electrodes that may be configured to sense electrical cardiac activity of the heart, to deliver electrical stimulation to the heart, or to sense patient thoracic impedance. In an embodiment, implantable lead system 110 may include one or more sensors configured to sense one or more other physiological parameters, such as cardiac chamber pressure or temperature. The conductive portion of the housing 101 (or connector) of the electronics unit 105 of the CRM device 102 may optionally act as an electrode, such as a "can" electrode.

For example, communication circuitry may be included within housing 101 (or a connector), such as for facilitating communication between the electronics unit 105 and an external communication device such as a portable or bedside communication station, patient-carried or patient-worn communication station, or external programming. communication between devices. Communication circuitry may also facilitate one-way or two-way communication with one or more implanted, mobile, external, cutaneous or subcutaneous physiological or non-physiological sensors, patient input devices, or information systems.

The CRM device 102 may include a motion detector 104 that may be used to sense patient physical activity or one or more respiratory or cardiac related conditions. In an embodiment, motion detector 104 may be configured to sense activity level or chest wall motion related to breathing effort. In an embodiment, motion detector 104 may include a single-axis or multi-axis (eg, three-axis) accelerometer, which may be located in or on housing 101 . Accelerometers can be used to provide useful information including information about the patient's position, respiratory information including, for example, about crackles or coughs, cardiac information including, for example, S1-S4 heart sounds, murmurs, or other sound information.

Processor circuitry 108 may be included within housing 101, for example. In an embodiment, the processor circuit 108 may include a plurality of data inputs configured to obtain physiological data from one or more physiological sensors. For example, processor circuit 108 may be configured to receive information from implantable lead system 110, eg, through an impedance measurement circuit connected to the first data input. In an embodiment, the first and second data inputs may be configured to receive information from the first and second physiological sensors, respectively. In an embodiment, the data input may be configured to receive chest impedance data measured using an electrode configuration defining a chest impedance vector. The third or fourth data input may be configured to receive information from the first and second physiological sensors.

Using one or more received physiological parameters, such as impedance values, processor circuit 108 may form a function. This function can be used as a basis on which multiple data clusters can be determined. Using the function, processor circuit 108 may determine at least one quantitative attribute associated with one or more data clusters. In an embodiment, the processor circuit 108 may be configured to determine an index of cardiac decompensation or a postural state, such as using one or more quantitative attributes. The processor circuit 108 may be configured to trend or otherwise monitor quantitative properties of one or more data clusters, such as using memory circuits.

In an embodiment, the processor circuit 108 may be configured to receive impedance-related information, such as using the implantable lead system 110 to receive voltage levels. Systems and methods describing obtaining impedance-related information are further described in Belalcazar, U.S. Patent No. 7,640,056 entitled "MONITORING FLUID IN A SUBJECT USING AN ELECTRODE CONFIGURATION PROVIDING NEGATIVE SENSITIVITY REGIONS )", which is incorporated herein by reference.

In an embodiment, a processor readable medium may be included, eg, within housing 101 . The processor-readable medium may include instructions that, when executed by a processor, configure the CRM device 102 to receive data, process data, interpret data, or provide data, such as using processor circuitry 108 . For example, a processor-readable medium may include instructions that, when executed by a processor, configure the CRM device 102 to form a function using impedance information received through a plurality of data inputs to the processor circuit 108 .

Storage circuitry may be included, such as within housing 101, for storing a number of values, including data trend information. In an embodiment, quantitative attributes of the data clusters, including extent along the y-axis, spread along the x-axis, area or volume, etc., may be stored in memory circuitry. In an embodiment, the storage circuitry may include a histogram-based storage mechanism to facilitate storage of quantitative attributes over extended periods of time. In embodiments, the storage circuitry may be external to the CRM device 102, or may be communicatively coupled to the CRM device 102 via communications circuitry.

The implantable lead system 110 and electronics unit 105 of the CRM device 102 may incorporate one or more thoracic impedance or similar signal sensors that may be used, for example, to obtain information about the patient's respiratory waveform or other respiratory correlations. information. Hartley et al., U.S. Patent No. 6,076,015 entitled "RATE ADAPTIVE CARDIAC RHYTHM MANAGEMENT DEVICE USING TRANSTHORACIC IMPEDANCE" describe methods for monitoring lung tidal volumes by measuring transthoracic impedance. Illustrative example, which is hereby incorporated by reference. In Hatlestad et al., U.S. Patent No. 7,603,170 entitled "CALIBRATION OF IMPEDANCE MONITORING OF RESPIRATORY VOLUMESUSING THORACIC D.C. IMPEDANCE" describes an illustrative example of a system that can detect respiratory signals and measure respiratory volume. example, which is hereby incorporated by reference.

In an embodiment, the thoracic impedance signal sensor may include, for example, one or more intracardiac electrodes 111 - 118 , such as may be positioned in one or more chambers of the heart 107 . Intracardiac electrodes 111 - 118 may be coupled to impedance drive/sensing circuitry 106 and may be positioned within the housing of pulse generator 105 , for example.

In an embodiment, the impedance driving/sensing circuit 106 may be configured to generate a current flow through tissue, such as between the impedance driving electrode 131 and a metal box electrode on the housing 101 of the electronics unit 105 . The voltage at the impedance sensing electrodes 114 relative to the metal box electrodes may vary as the patient's chest impedance varies. The voltage signal developed between the impedance sensing electrode 114 and the metal box electrode may be detected by the impedance sensing circuit 106 . Other positioning or combinations of impedance sensing or driving electrodes are also possible. Some examples are listed in Table 1 and discussed below.

Implantable lead system 110 may include one or more cardiac pacing/sensing electrodes 113-117, such as may be positioned in, on, or around one or more heart chambers, e.g., for sensing One or more electrical signals from the patient's heart 107. Intracardiac sensing and pacing electrodes 113-117, such as those shown in FIG. Left atrium (LA) or right atrium (RA). The implantable lead system 110 may include one or more defibrillation electrodes 111, 112, such as for delivering a defibrillation or cardioversion shock to the heart or for sensing one or more intrinsic electrical signals from the heart 107 .

2A and 2B illustrate generally embodiments of impedance data such as may be acquired using implantable lead system 110. In the embodiment of FIG. 2A , graph 200 illustrates an example of chest impedance data such as may be measured using right atrial (RA) electrodes, such as RA impedance sensing electrodes 114 and metal box electrodes. In an embodiment, the RA-Can electrode configuration may indicate a first impedance vector. In the embodiment of FIG. 2B , graph 220 illustrates an example of chest impedance data that may be measured, for example, using right ventricle (RV) electrodes such as RV sense electrodes 115 and metal box electrodes. The RV-Can electrode configuration can indicate a second impedance vector.

Thoracic impedance data may be obtained instantaneously from the patient at a particular moment, or may be recorded and monitored over a period of time, such as over a day, days, or a longer period of time. In the embodiment of FIGS. 2A and 2B , the data shown in graphs 200 and 220 may represent thoracic impedance data, such as may be obtained from a patient over a period of about 14 days. In an embodiment, chest impedance data may be obtained at a sampling rate of approximately 72 samples per day, or 1 sample every 20 minutes, for each monitored impedance vector. Each peak-to-peak period of measured impedance data can represent approximately one 24-hour period, depending on the patient's activity and physiological parameters. For example, the time period defined by peak 203 and peak 204 may represent a 24 hour period. The data represented in graph 200 can be obtained during the same 14-day period as the data represented in graph 240 below, such that the peak of the RA-Can impedance measurement coincides in time with the peak of the RV-Can impedance measurement (e.g., peak 203 occurs at about the same time as peak 223, eg, within the same sampling time interval, and peak 204 occurs at about the same time as peak 224).

In one embodiment, the data points represented by the blackened circles in FIGS. 2A and 2B may represent impedance measurements, such as may be obtained while the patient is in a lateral or supine position. Data points represented by open circles in FIGS. 2A and 2B may represent impedance measurements, such as may be obtained when the patient is in an upright position. While body position can be roughly inferred based on time of day, a more precise body position detection mechanism can be obtained using the impedance information represented in Figures 2A and 2B, as explained below. Useful trend information, such as that used to provide an indication of decompensation in heart failure, can also be obtained using the impedance information in Figures 2A and 2B.

FIG. 3 generally illustrates an embodiment of a graph 300 illustrating on a common axis an embodiment of recorded impedance from RA-Can and RV-Can impedance vectors, such as during a period comprising about four days. Graph 300 includes inclination data from a body position sensor, such as motion detector 104 . In the embodiment of FIG. 3 , a 0 degree inclination may represent a lateral, supine or supine patient position or position, and a 90 degree inclination may represent an upright or standing patient position or position.

FIG. 3 generally illustrates examples of fluctuations in thoracic impedance that may be expected, for example, due to patient movement, neurohormonal modulation, or other factors affecting the patient. Typically, changes in thoracic impedance over time are also reflected in more than one measured thoracic impedance vector. In the example of Figure 3, it can be seen that the magnitude of the RA-Can impedance vector 331 from one sample to the next varies by approximately the same magnitude as the RV-Can impedance vector 332 of the temporally corresponding sample. For example, the change in impedance of RA-Can impedance vector 331 from a sample before _t2 to a sample after _t2 is _Δ2A , or about 2 ohms. The change in impedance of the RV-Can impedance vector 332 from the sample before t ₂ to the sample after t ₂ is Δ _2V , which is about 2 ohms. Thus, Δ _2A is approximately equal to Δ _2V , and the magnitudes of the two impedance vectors are adjusted by approximately the same amount. Similarly, the change in impedance for the RA-Can impedance vector 331 from the sample before t4 to the sample after t4 is _Δ4A , and for the RV _- Can impedance vector 332 is _Δ4V , where _Δ4A is approximately equal _{to Δ4V .} In other words, while the RA-Can and RV-Can impedance vectors are expected to vary throughout a given period of time, eg, in response to normal circadian rhythm changes, the magnitudes of both vectors can be expected to vary by approximately the same amount. That is, generally, (│RA-Can _t -RA-Can _t-1 │)≈(│RV-Can _t -RV-Can _t-1 │).

However, at times _t1 and _t3 , the change in magnitude of the impedance of the RV-Can impedance vector 332 may be greater than the magnitude of the change in impedance of the RA-Can impedance vector 331 . In the example of FIG. 3, time _t1 occurs just before midnight, and the body position information or inclination indicates that the patient has undergone a body position change from upright to side lying, such as when the patient enters a night's sleep. Following this change in body position, changes in the magnitude of the impedance vector can be observed. It is believed that the change in magnitude of the impedance vector upon body position change can be attributed, among other factors, to the sudden shift of the patient's pleural fluid.

In the embodiment of FIG. 3 , the magnitude change of RA-Can impedance vector 331 is Δ _1A , or about 1 ohm, and the magnitude change of RV-Can impedance vector 332 is Δ _1V , or about 4 ohms. Therefore, a larger magnitude change of the RV-Can impedance vector 332 compared to the RA-Can impedance vector 331 can be observed when the patient position changes. Similarly, at _t3 , the patient undergoes a second postural change from a lateral to an upright position, such as when the patient wakes up during the day. The change in impedance from the sample before _t3 to the sample after _t3 is _Δ3A for RA-Can impedance vector 331 and _Δ3V for RV-Can impedance vector 332 . Likewise, the magnitude change _Δ3V of the RV-Can impedance vector 332 is larger than the magnitude change _Δ3A of the RA-Can impedance vector 331.

FIG. 4A generally illustrates an embodiment of a graph 400 using a Cartesian coordinate system where information from a first physiological sensor can be plotted against information from a second physiological sensor. In the embodiment of FIG. 4A , impedance data from RA-Can impedance vector 331 (eg, along the x-axis) can be plotted against impedance data from RV-Can impedance vector 332 (eg, along the y-axis). Impedance magnitude samples taken at corresponding times or within corresponding sample windows form the coordinates of the plotted impedance data.

In the embodiment of FIG. 4A , the first time interval 0<t<1 may include impedance magnitude measurements for each of the RA-Can impedance vector 331 and the RV-Can impedance vector 332 . The magnitude of the first impedance data obtained during the first time interval is plotted against the second impedance data obtained during the same first time interval. Additional impedance magnitude data may be added for subsequent time intervals, eg, every half hour.

FIG. 4A is an example illustrating about 4 days of impedance information from a single patient, such as using the impedance information shown in FIG. 3 . In one embodiment, the impedance data shown in Figure 4A may be sampled at a rate of one sample every 20 minutes, for a total of 288 data points during the four day period. In Figure 4A, the impedance data from the RA-Can vector and the RV-Can vector can be plotted on a Cartesian plane, forming at least two data clusters. Data clustering may occur due to differences in the magnitude change (eg Δ _1A ) of the first vector compared to the second vector (eg Δ _1V ) as the patient position changes. A data cluster can be a subset of the original data. The partitioning or assignment of raw data into data clusters can be accomplished using one or more of several clustering techniques, eg, using a computer processor to execute instructions to perform the clustering techniques. A variety of clustering techniques can be used, including hierarchical clustering (e.g. involving the division of a large data set into successively smaller clusters), partitional clustering (e.g. involving the determination of an attribution factor for each point to each cluster), or Density-based clustering (e.g., involves identifying high-density regions). Each cluster may represent a portion of a function formed using information from two or more physiological sensors.

Referring now to FIGS. 3 and 4A , impedance data for RA-Can impedance vector 331 and RV-Can impedance vector 332 for t < t ₁ may be represented in lower cluster 542 of FIG. 4A . At time t=t ₁ , the magnitude change _Δ1V of the RV-Can impedance vector 332 is greater than the magnitude change _Δ1A of the RA-Can impedance vector 331 . This difference in amplitude change may result in a jump from the lower cluster 442 corresponding to the upright patient position to the upper cluster 441 corresponding to the side lying patient position. Impedance data acquired during times t ₁ <t<t ₂ , such as corresponding to a lateral patient position, may fall within the upper cluster 441 . At time t= _t3 , a change in patient position, such as returning the patient to an upright position, may result in another jump. A transition may be from upper cluster 441 to lower cluster 442 . Impedance data acquired for t> _t3 may fall within the lower cluster 442, corresponding to the upright patient position.

In an embodiment, graph 400 may include first physiological data obtained using a first physiological sensor, and second physiological data obtained using a second physiological sensor. In an embodiment, the first and second physiological data may be acquired over time, and the time variation may be removed from the graph 400 by plotting the first physiological data against the second physiological data.

Physiological sensors may include sensors configured to measure electrical properties such as voltage or impedance, or sensors configured to measure mechanical or acoustic information, among others. In an embodiment, three or more physiological sensors, such as three different implantable electrodes, may be used to monitor three different impedance vectors in the patient. In an embodiment, two physiological sensors may be configured to monitor two different impedance vectors, and a third physiological sensor may be configured to monitor the S3 heart sound. Data from the two or more physiological sensors may be used to form a function, such as a function of the first physiological data on the second or third physiological data. Functions can be used to form clusters of data, such as by determining discrete parts of the function, such as by plotting the data in a time-independent manner.

In an embodiment, map 400 may include a third impedance vector, such as using left ventricular electrodes and metal box electrodes. This third impedance vector can be plotted against the RA-Can impedance vector 331 and the RV-Can impedance vector 332 to form a three-dimensional data cluster. Additional sensor data can be plotted against the three-dimensional data cluster, for example, to form a multidimensional function.

FIG. 4B generally illustrates several of the characteristic or quantitative attributes of graph 400 . Some examples of properties may include spread along the x-axis, extent along the y-axis, and data cluster centroid locations. Some properties of multidimensional functions, for example, can be formed using data from two or more sensors, and can include spread, range, area, volume, and hyper-volume (in the case of 4-dimensional or higher-dimensional spaces), and other properties.

In an embodiment, the spread 450 of one or both of the data clusters 441, 442, such as along the x-axis, may be measured. The expansion 450 of the data cluster may represent a domain of all impedance magnitude values attributed to a particular vector, such as the RA-Can impedance vector 331 . In the embodiment shown in FIG. 4B, the extension of the lower cluster 442 may be 33 ohms to 46 ohms. In an embodiment, the difference between the maximum impedance and the minimum impedance may represent the extension 450 . In the embodiment shown in FIG. 4B, the extension 450 of the lower cluster 442 may be 13 ohms and the extension 450 of the upper cluster 441 may be 16 ohms. In the embodiment shown in FIG. 4B , the expansion 450 of the upper cluster 441 may be equal to the expansion 450 of the entire function represented on the graph 400 . In an embodiment, the spread 450 may represent the standard deviation or variance of the impedance values of the data points within the cluster.

FIG. 4B generally illustrates an embodiment of the range 451 of the function represented on the graph 400 . Range 451 may represent the range of the entire function represented on graph 400, or range 451 may represent a portion of a function, such as a range that includes one or more clusters of data. In the embodiment of FIG. 4B, the range 451 of the upper cluster 441 may be about 15 ohms. In an embodiment, the range of the lower cluster 442 may be about 13 ohms. In an embodiment, the range 451 may represent the standard deviation or variance of the impedance values of the data points within the cluster.

In an embodiment, the scope of the function and the expansion can be combined, such as using the Pythagorean theorem, to obtain the scope of the function on a multi-dimensional space. For example, using the equation Range and Spread of the function can be combined, where Spread(x) is the full spread of the data along the x-axis, and Range(y) is the full range of the data along the y-axis.

Quantitative properties of the data, such as centroids of data clusters, distances between data cluster centroids, or areas of one or more data clusters, among other properties, can also be measured using graph 400 . FIG. 4B generally illustrates the centroid 452 of the upper cluster 441 and the centroid 453 of the lower cluster 442 . For example, the centroid of the upper cluster 441 may be located at the coordinates:

$<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; RA</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; can</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; RA</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; can</span>}{<span\; class="notranslate">\; )</span>}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; RV</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; can</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; RV</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; can</span>}{<span\; class="notranslate">\; )</span>}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>$

where n is the number of RA-Can impedance measurements attributed to the upper cluster 441 and (RA-Can) _i is the magnitude of the impedance measurement attributed to sample i.

The distance between the centroids of the upper cluster 441 and the lower cluster 442 can be determined, for example, using the Pythagorean theorem. In general, centroid analysis can be extended to any number of clusters, including clusters in three-dimensional space, for example, can be plotted using data from additional sensors or impedance vectors. The centroid coordinates in the third dimension can be:

where n is the number of measurements from the third sensor attributed to the upper cluster 441 and (sensor) _i is the magnitude of the measurement from the third sensor attributed to sample i.

In an embodiment, analyzing one or more vectors associated with a centroid may provide quantitative attribute information. For example, a vector related to the origin and centroid 452 may be compared to a vector related to the origin and centroid 453 . In an embodiment, a dot product of two or more vectors may be performed to obtain quantitative attributes, including angles between vectors.

There are several ways to calculate the area of a data cluster, such as by integrating a function or set of functions representing the data, or by numerical integration techniques, etc. Figure 4C generally illustrates a method of calculating the area of a data cluster using numerical integration. The spread or range of data can be divided into n discrete intervals, and local minima and maxima can be found for each interval to define a rectangular extent. The rectangular areas can be determined and summed to provide an approximation of the total cluster area. In the embodiment of FIG. 4C, the data clusters may each be divided into 14 bins of 1 ohm each. The data points in interval 6 of the upper cluster 441 include a local minimum 455 at (RA-Can=36.3, RV-Can=33), and a local minimum at (RA-Can=36.25, RV-Can=36.5). The maximum value is 454 to define a rectangle with an area of 3.5 (units omitted). The area of the rectangle in the 6th interval of the lower cluster is 2.25 (units omitted). The total area calculated using this method is about 29.25 for the upper cluster and about 22.5 for the lower cluster.

The area of the data cluster group can also be calculated, as shown in Figure 4D. A rectangle can be drawn that completely encloses the cluster of data. In the example of FIG. 4D , the total area of the rectangle enclosing the data clusters is about 74 . Several other techniques for calculating or approximating the total area of data clusters can be used, such as the numerical integration method described above in the discussion of Figure 4C.

In an embodiment, the area of a pair of data clusters can be calculated using the following equation:

where d is the distance between the centroids of the two clusters, spread (x) is the full spread of the data along the x-axis, and extent (y) is the full extent of the data along the y-axis. In an embodiment, it is assumed that data clusters can be represented as having length A quadrilateral with two opposite sides, and the distance between the two opposite sides is equal to the distance between the centroids.

Other data clustering analysis techniques may be used. For example, least squares techniques can be used to regress a line of best fit to one or more data clusters. The slope of the line of best fit can be used as a quantitative attribute, or the area under a portion of the line of best fit can be used. Other techniques for regressing higher order best fit curves to one or more data clusters may also be used.

Data cluster density can be monitored over time, such as over a particular range or spread. Density of change or location of density can be used as a quantitative attribute to provide diagnostic information for a patient. In an embodiment, a function representing the density may be formed and analyzed to determine a quantitative attribute.

4A-4D generally illustrate non-limiting examples of several quantitative attributes of data clusters 441 and 442 . These examples should not be viewed as limiting the scope of the present subject matter. There are many ways of computing quantitative properties associated with clustering of data.

Various methods for acquiring and representing physiological data have been described with respect to a first patient physiology (eg, a physiology that is currently not indicative of cardiac decompensation). In a second patient physiology, such as a diseased physiology, the physiology data may be substantially different from the first physiology. For example, the magnitude of modulation of a particular thoracic impedance vector may differ substantially between a first patient physiology and a second patient physiology.

FIG. 5 generally illustrates an embodiment of a graph 500 illustrating an embodiment of impedance vector and patient position data from a patient at increased risk of heart failure decompensation. Diagram 500 illustrates a period comprising approximately 4 days of recording positional information and recording impedance data for the patient's RA-Can impedance vector 531 and RV-Can impedance vector 532 . In the embodiment of FIG. 5, the RV-Can impedance vector 532 does not exhibit a significant change in magnitude when approaching or at a change in patient position.

As in FIG. 3 , FIG. 5 illustrates fluctuations in thoracic impedance that can be expected due to patient movement, neurohormonal modulation, or other factors affecting the patient. However, the impedance vector depicted in FIG. 5 may be indicative of a diseased patient state, such as a patient experiencing pulmonary edema. In particular, the difference in magnitude change of the RA-Can impedance vector 531 compared to the RV-Can impedance vector 532 may be reduced or minimized as the patient position changes.

For example, at t ₅ and t ₇ , the change in impedance magnitude of RV-Can impedance vector 532 no longer contrasts significantly with the change in impedance magnitude of RA-Can impedance vector 531 . In the embodiment of FIG. ₅ , time t5 occurs just before midnight, and the position information indicates that the patient has undergone a change from an upright to side-lying position, such as when the patient enters a night's sleep. Following this body position change, the large change in RV-Can impedance vector 332, as observed at _Δ1V in FIG. 3, is reduced or no longer exists.

In an embodiment, changes in the magnitude of the RA-Can or RV-Can impedance vector as the patient changes position may be due, among other factors, to local shifts in the patient's pleural fluid. Pleural fluid in an upright patient may tend to accumulate in the lower regions of the body due to the normal response of the fluid due to gravity. Fluid can be dispersed in the chest, including the heart region, when the patient is placed in a lateral or supine position. Fluid accumulation in the region where the electrodes are implanted can contribute to the impedance changes detected using these electrodes, as can be the nocturnal increase in thoracic admittance due to fluid accumulation in the lungs. Patients experiencing pulmonary edema, as can be caused by venous congestion associated with congestive heart failure, may develop abnormal accumulation of pleural fluid. However, the magnitude of normal fluid transfer between upright and lateral positions is reduced as fluid accumulates. That is, because fluid levels may be elevated in disease states, fluid shifts may not be as great when the patient changes position. Changes in the magnitude of the measured thoracic impedance vector due to changes in patient position can be reduced if the lung region is saturated with fluid. In some vectors, fluid transfer can be reflected more easily than in others.

In the embodiment of FIG. ₅ , the magnitude change of RA-Can impedance vector 531 at t5 is _Δ5A , or about 1 ohm, and the magnitude change of RV-Can impedance vector 532 is _Δ5V , or about 1 ohm. The change in magnitude of the RV _- Can impedance vector 532 at t5 is significantly less than the change in magnitude of the RV-Can impedance vector 332 at _t1 . This substantial difference at similar postural changes may indicate increased pleural fluid levels. Likewise, at t ₇ , the patient undergoes a second change in position, from side lying to an upright position. From the sample before t7 to the sample after t7, the impedance magnitude changes by _Δ7A for RA _- Can impedance vector 531 and _Δ7V for RV-Can impedance vector ₅₃₂ . Importantly, the magnitude change of the RV-Can impedance vector 532 , Δ _7V , is less than the magnitude change of the RV-Can impedance vector 332 , Δ _3V .

FIG. 6 generally illustrates an embodiment of a second graph 600 in which impedance data from RA-Can impedance vector 531 can be plotted against corresponding impedance data from RV-Can impedance vector 532 . As in Figure 4A, when two vectors of impedance data are plotted against each other to remove time as the plot axis variable, the data can form clusters. However, in the embodiment of Figure 6, the data may indicate elevated fluid levels, such as pulmonary edema due to worsening heart failure. Comparing FIGS. 6 and 4A , the clusters in the second graph 600 are not as clearly distinguishable as they are on graph 400 . Since the magnitude of the change difference in the first vector (eg Δ _5V ) is reduced compared to the magnitude of the change difference in the second vector (eg Δ _5A ) when the patient position changes, in RA-Can/RV-Can Cartesian Clusters are close to each other in the plane. In an embodiment, cluster boundaries may become indistinguishable. An upper cluster 641 representing a first body position may be indistinguishable from a lower cluster 642 representing a different body position.

In the example of FIG. 6, the impedance modulation of the vector due to normal circadian rhythm changes over a 4-day period can be reduced in magnitude compared to the modulation observed in the vector during different time periods, such as including the healthy, non-edematous state. Small. For example, the magnitude of the impedance change at _Δ8A and _Δ8V on the sixth day may be lower than the magnitude of the impedance change at the same time on the second day, such as Δ4A and _Δ4V in FIG. _4A . Changes in the magnitude of impedance changes may indicate different patient physiological states. The reduced impedance modulation of FIG. 6 , as compared to FIG. 4A , can be observed in the spread and extent of the data clusters in the second graph 600 . Spread and range are each reduced from the different physiological states represented in graph 400 . For example, the spread of upper cluster 441 is 17 ohms, and the spread of upper cluster 641 is 10 ohms.

Quantitative properties of data clusters formed using recently acquired data may be compared to properties of data clusters formed using previously acquired data, such as detecting changes in patient physiology. Such changes can be monitored by monitoring quantitative properties of data clusters associated with several discrete boundary bundles of physiological data, such as weekly or daily physiological data. Quantitative properties of the data clusters, in an embodiment, may be monitored or trended over an arbitrary period of time using the processor circuit 108, eg, to determine one or more trends in patient etiology. For example, data clustering analysis over time can be used to indicate diseased or at-risk patient status. This can include providing indicators of heart failure decompensation. Data cluster attribute information can be stored in any number of formats, for example, in the format of a histogram to facilitate storage of large amounts of data.

One or more quantitative attributes of data clusters such as area, centroid location, spread or volume, etc. can be analyzed and trended. In an embodiment, the area of one or both data clusters in second graph 600 may be compared to the area of one or both data clusters in graph 400 . In an embodiment, upper cluster 641 may be compared to lower cluster 441 . The area of one or two data clusters in the second graph 600 may be compared to a trend of cluster areas including more than one previous cluster area. Where three physiological sensors are used to obtain the three-dimensional function, the volume of the data clusters can be trended. In an embodiment, more than four physiological sensors may be used to obtain multidimensional functions. Quantitative properties of clusters defined by multidimensional functions can be trended. In an embodiment, trending one or more quantitative attributes may include monitoring and recording quantitative attributes for a series of data clusters obtained during several time windows, such as consecutive time windows.

The centroid locations of the data clusters 641, 642 may be compared to the centroid locations of previously recorded data. For example, the location of centroid 652 may be compared to the location of centroid 452 . In an embodiment, a change in the centroid position of a data cluster associated with a particular patient's position may indicate a physiological trend, such as an increased risk of cardiac decompensation. The location of the centroid 652 may be compared to a series of previous centroid locations associated with the upper data cluster, for example, to discover trends in the upper data cluster centroid locations.

The distance between two or more centroids can be recorded over a period of time and trended, eg, to indicate one or more changes in the physiological state of the patient. A first distance between centroids 452 and 453 may be determined, such as using the Pythagorean theorem. The second distance between centroids 652 and 653 can be similarly determined. The distance between any number of centroids representing clusters of data reflecting several body positions can be calculated and trended. In an embodiment, a decrease in the distance between the first and second distances may indicate an abnormal patient condition, such as an existing or impending decompensation of heart failure associated with pulmonary edema.

The maximum and minimum values of each data cluster can be trended, or the extent or spread of the data clusters can be trended over time. For example, the spread 450 of the upper data cluster 441 can be compared to the spread 650 of the upper data cluster 641, or it can be compared to the trend of the spread of a series of upper data clusters. The expansion 450 of the function represented in graph 400 may be compared to the expansion 450 of the function represented in the second graph 600 , or the trend of the expansion of several functions. In an embodiment, a reduced spread in the second graph 600 compared to graph 400 may be indicative of a patient physiological change.

In an embodiment, a baseline physiological patient state may be established, eg, including quantitative attributes of the baseline data cluster. The baseline data cluster quantitative attributes may include a baseline distance between data cluster centroids of different data clusters representing upright and side lying positions. In an embodiment, a change in distance from baseline beyond some threshold distance may send an alert to the patient or physician to provide an early indication of a potentially abnormal condition.

7 generally illustrates an embodiment 700, which may include obtaining first physiological data 712, obtaining second physiological data 714, forming a function 720 of the first physiological data to the second physiological data, forming at least two data clusters 722, determining the data Clustering attributes 762, and using the data clustering attributes to provide heart failure decompensation indicators 782.

At 712, first physiological data can be obtained. At 714, second physiological data can be obtained. Obtaining physiological data may include obtaining one or more of data indicative of an electrical characteristic of the patient, data indicative of a mechanical characteristic of the patient, or data indicative of a current state of the patient. Data indicative of electrical properties may include information indicative of impedance, intrinsic tissue signal, capacitance, or admittance, among other types of information. Data indicative of mechanical properties may include information indicative of patient movement, acoustic information including heart sounds, or respiration information, among other types of information. Data indicative of the current state of the patient may include information about the patient's position or the patient's activity level, among other types of information.

In an embodiment, obtaining first physiological data 712 may include obtaining chest impedance measurements using electrodes positioned on or in the patient's body. Implantable electrodes installed in or near the heart wall, such as right ventricle sensing electrodes 115, may be used. A second electrode, such as a metal can (Can) electrode, may be positioned elsewhere, for example in the patient's chest or abdomen. A first impedance vector, eg, between the right ventricle sensing electrode 115 and the metal box electrode, may be used to obtain chest impedance measurements. Several other vectors that can be used to obtain physiological data are listed in Table 1, where "X" indicates a possible vector between the corresponding first and second electrodes. In Table 1, Can refers to the conductive housing of the implantable device, RA refers to one or more electrodes positioned in or near the right atrium of the heart, and RV refers to electrodes positioned in or near the right ventricle of the heart. One or more electrodes, LV refers to one or more electrodes positioned in or near the left ventricle of the heart, SV refers to one or more electrodes positioned in or near the supraventricular region of the heart, connector plug block refers to is a lead connector plug block (e.g., "header") electrode, as on the CRM102, which is separate from the metal case electrode, and intravascular cathode refers to one located at an intravascular location other than the heart or multiple electrodes. In an embodiment, obtaining thoracic impedance measurements may include averaging or otherwise calculating a central tendency of all measured impedances such that changes in impedance due to cardiac beating or respiration are largely omitted.

Table 1 Impedance vectors used to obtain physiological data.

In an embodiment, obtaining first physiological data 712 may include obtaining heart sound information using a sensor such as an accelerometer configured to detect S3 heart sound information. The first physiological data may include electrical signals from an accelerometer indicative of mechanical vibrations of the heart.

At 720, a function can be formed using the first physiological data and the second physiological data. In an embodiment, the first physiological data may be used as a range of values, and the second physiological data may be used as a range of values. Each value in the domain may correspond to a value in the range, for example by pairing values obtained at the same time, or during a common time interval or window. In an embodiment, the first and second physiological data may be plotted against each other on a common axis. In case the first and second physiological data are obtained in a time-dependent manner, the plotting of the first physiological data against the second physiological data may remove time as an explicit variable in the formed function.

At 722, one or more data clusters can be formed. In an embodiment, data clusters may be obtained from the function formed at 720 . One or more of several clustering techniques may be used, such as described above in the discussion of FIG. 4A. In an embodiment, two impedance vectors may be used to form the function, and at least two data clusters may be discerned from the function. If desired, the number of data clusters can be limited, for example, by the clustering technique or by available processor capacity. In an embodiment, only a single cluster can be discerned from the function using available clustering techniques.

In embodiments comprising two clusters of data obtained from the function using two impedance vectors, one cluster may correspond to the build-up phase, or build-up phase, during which impedance is measured throughout the waking portion of the patient's day. During the ramp-up phase, the impedance data may correspond to a first data cluster, and during the ramp-down phase, the impedance data may correspond to a different data cluster. Depending on patient position, impedance may tend to be higher or lower in average magnitude, and may tend to grow and fall over time in an approximately periodic or other recurring fashion, as illustrated in Figures 2, 3 and 5 .

At 762, one or more data cluster attributes can be determined, such as spread, extent, mean, centroid location, area, density, or volume. Several non-limiting methods for determining data cluster properties are included in the discussion of Figures 4A, 4B, 4C and 4D.

At 782, a heart failure decompensation index can be provided using one or more data cluster attributes. In an embodiment, any one or more quantitative properties, such as the spread, extent, mean, centroid location, distance between two or more centroids, area, density, or volume of a data cluster, alone or in combination Therefore, it can be used to provide indicators of decompensation in heart failure. In an embodiment, a comparison of one or more quantitative attributes associated with a first time window with one or more quantitative attributes associated with a different time window may be used to provide an index of heart failure decompensation. Below, the discussion of FIG. 8 includes several examples of using data cluster attributes to determine abnormal or disease states, such as determining heart failure decompensation indicators.

8 generally illustrates an embodiment 800 that may include obtaining first and second physiological data 810 during a first time window, forming a first function 820 of the first physiological data versus second physiological data, using the first function to determine a first quantitative attribute 830, obtaining third and fourth physiological data 840 during a second time window, forming a second function 850 of the third physiological data on the fourth physiological data, determining a second quantitative attribute 860 using the second function, evaluating the first The first and second quantitative attributes are trended 870 and the trends are used to provide a heart failure decompensation index 880 .

At 810, first and second physiological data can be obtained during a first time window. Obtaining physiological data may include obtaining one or more of data indicative of an electrical characteristic of the patient, data indicative of a mechanical characteristic of the patient, or data indicative of a current patient state, wherein the physiological data may be obtained during discrete time intervals. A first physiological sensor may be used to obtain first physiological data, and a second physiological sensor may be used to obtain second physiological data. The first and second physiological sensors may be electrodes in implantable lead system 110 .

In an embodiment, the first and second physiological data may be obtained at any instant or at multiple instances during the first time window, such as during the first 20 minute window. For example, the first and second physiological data may be obtained by averaging or calculating another central tendency of a series of measurements in a 20 minute window, or by storing the maximum or minimum measurements occurring in the 20 minute window. The window duration and number of data collection instances during the window may be any suitable duration or number that will allow accurate collection of physiological data. For example, impedance measurements may be taken using the implantable lead system 110 as an average or other central tendency of the impedance over a period of milliseconds. Measurements to detect heart wall motion can involve windows of several seconds, such as detecting muscle responses to electrical stimulation.

At 820, a first function of the first physiological data can be formed for the second physiological data. The first function may be formed, eg, according to the discussion of 720 of FIG. 7 . In an embodiment, part of the first function may determine at least a first pair of data clusters.

At 830, a first quantitative attribute can be determined using a first function. The first quantitative property may include any property of the first function, such as including the range, domain, or periodicity of the function, among other properties. The first quantitative attribute may be from one or more data clusters formed using one or more parts of the function, such as the first pair of data clusters. For example, a function may describe multiple data clusters, and the first quantitative attribute may include the area of a particular data cluster.

At 840, third and fourth physiological data can be obtained during a second time window. The third and fourth physiological data may be obtained in the same manner as the first and second physiological data, although the third and fourth physiological data may correspond to different time intervals. In an embodiment, the first and third physiological data may be obtained using the same physiological sensor, such as an accelerometer configured to obtain heart sound information. The second and fourth physiological data may similarly be obtained using the same physiological sensor, such as an implantable electrode positioned on the patient's chest. In an embodiment, the second time window may be of a different duration than the first time window. Different time window durations or sampling rates can be used to increase or decrease data collection at specific times. For example, an increased sampling rate may be used when a specific event is expected, such as a patient position change. A reduced sampling rate can be used, for example, when the patient is sleeping or immobile.

At 850, a second function can be formed using the third and fourth physiological data. The second function may be formed in the same manner as the first function, as described above in the discussion of 820 of FIG. 8 . In an embodiment, part of the second function may determine at least a second pair of data clusters.

At 860, a second quantitative attribute, such as a test quantitative attribute, can be determined using a second function. The second quantitative property may include any property of the second function, such as including properties described at 830 in this discussion. In an embodiment, the second quantitative attribute may be from a second pair of data clusters.

At 870, first and second quantitative attributes, such as discerning trends, can be evaluated. In an embodiment, the first and second quantitative properties may indicate the total area of the first and second functions, respectively. The evaluation may indicate that the total area decreases from the first time window to the second time window. In an embodiment, the first time window may indicate patient chest impedance measurements taken during a first week, and the second time window may indicate chest impedance measurements taken during a subsequent week. In an embodiment, the evaluation of the first and second quantitative attributes of the total area may indicate a decreasing area trend. Additional quantitative attributes can be evaluated to determine trends over a longer period of time, such as weeks or months. Trends, such as a decrease in total area, may indicate patient health status or risk factors.

At 880, trends can be used to provide indicators of heart failure decompensation. In an embodiment, the total area defined by the first and second functions may represent a decreasing trend. A reduced total area may indicate that the patient is at increased risk of heart failure decompensation. An alert may be provided to the patient or clinician upon detection of increased patient risk, such as in the CRM 102 using telemetry circuitry that is communicatively coupled to an external patient management device.

Turning now to position detection, patient position can affect patient thoracic impedance. Sensors, such as comprising electrode pairs comprising a metal box electrode and one or more electrodes from the implantable lead system 110, can detect the patient's position or changes in position. The electrodes may be positioned on opposite sides of the patient's chest, or may be positioned in cardiac tissue, or a combination of such positioning. For example, electrodes may be positioned near a patient's spine to deliver therapy, and an implantable device having electrodes included at its housing 101 may be implanted in the patient's abdomen or chest.

FIG. 9 generally illustrates an embodiment 900 that may include obtaining a postural discriminative metric 905 , obtaining thoracic impedance test data 915 , comparing 971 the thoracic impedance test data to the postural discriminative metric, and providing a postural status 991 .

At 905, a body position discriminant metric can be obtained. In an embodiment, obtaining the body position discriminant metric may include using the first chest impedance data and the second chest impedance data. The first chest impedance data may be measured using a first electrode configuration defining a first chest impedance vector, such as using a combination of electrodes in implantable lead system 110. The first chest impedance data may correspond to one or more instances during the first time window. Second thoracic impedance data may be measured using a second electrode configuration defining a second thoracic impedance vector, wherein the second electrode configuration is different from the first electrode configuration. The second chest impedance data may correspond to one or more instances during the same first time window.

In an embodiment, quantitative attributes of the first chest impedance data and the second chest impedance data may be used to form a postural discriminant metric. The mean or median or other central tendency of a set of impedance data can be used, such as to establish a threshold impedance. Impedance values above a threshold may correspond to a first body position, and impedance values below the threshold may correspond to a second body position. In an embodiment, several quantitative attributes may be used to form a body position discriminative metric, eg, to improve the specificity of the metric. Instead of using the mean or median of a set of impedance data, the magnitude of the impedance vector modulation during a given time window can be used, eg, to create a more discriminative metric.

In an embodiment, the first chest impedance data and the second chest impedance data may be used to form a function, such as may be obtained by plotting the first chest impedance data against the second chest impedance data on a rectangular Cartesian coordinate. This function can be discretized as a function of one or more parts to form a body position discriminant metric. For example, a first part of the function may correspond to a first body position, and a second part of the function may correspond to a second body position. In an embodiment, a portion of a function may be associated with a data cluster, or several different portions of a function may be associated with several different data clusters.

A data cluster formed using the first chest impedance data and the second chest impedance data may be used to form a postural discriminant measure. The discrete data clusters may correspond to multiple patient positions and may be used to determine patient position, such as by comparing currently acquired physiological information. In an embodiment including two data clusters, the first data cluster may represent a first patient position and the second data cluster may represent a second patient position. Additional data clusters, if present, may represent intermediate patient positions. Various techniques, such as regression analysis or discriminant analysis, can identify the boundaries of data clusters.

In an embodiment, obtaining a body position discriminative metric may include a learning or training period. The first and second chest impedance data may be monitored, such as during a first time window including one week, to establish a baseline patient impedance data set. Clustering of the baseline dataset can be assessed, and body positions can be attributed to specific clusters. Position-cluster correlations can be established manually, as by a clinician, or automatically, as by processor-driven analysis. In an embodiment, a learning period may be performed in a clinical setting during or after implantation of the CRM 102 . In an embodiment, the learning period may include the use of a body position sensor or an accelerometer.

In an embodiment, the time-of-day information may be used at least in part to determine the body position comparison metric. Changes in the patient's position can generally be expected at regular periodic intervals, such as including when the patient will be lying down at night or rising in the morning. Thus, when evaluating the accumulation of thoracic impedance data to establish a body position comparison metric, the processor or clinician can look at a relatively narrow window of time in which body position changes may have occurred. In an embodiment, the patient may record sleep/wake times during a first time window for subsequent correlation with impedance data acquired during the same first time window.

At 915, chest impedance test data can be obtained. Thoracic impedance measurement data can be obtained using a variety of hardware components, including implantable lead system 110, subcutaneous electrode arrays, or electrode arrays positioned on the body surface. In an embodiment, at least two electrodes are used to define a test impedance vector for obtaining chest impedance test data. In an embodiment, obtaining chest impedance test data may include using the same implantable lead system 110 or the same portion of the same implantable lead system 110 that may be used during a study or training session. The vector configuration used to obtain the chest impedance test data at 915 may be substantially the same as the vector configuration used during the learning or training period.

At 971, chest impedance test data can be compared to a body position discriminant measure. The comparison of the impedance test data to the body position discriminant metric may be performed using the processor circuit 108 . Comparisons can also be performed manually, for example by clinicians evaluating patient data. In an embodiment, chest impedance test data may be obtained using a test impedance vector and may be compared to the one or more postural comparison metrics obtained at 905 . In an embodiment, wherein the body position comparison metric comprises an average of the first chest impedance data and the second chest impedance data, the chest impedance test data may be compared to the average.

In an embodiment, the chest impedance test data may include data from the first test impedance vector and the second test impedance vector. The first chest impedance vector may be the same as the first test impedance vector, and the second chest impedance vector may be the same as the second test impedance vector. Where data from the first chest impedance vector and the second chest impedance vector (which includes at least two data clusters) are used to form the function, the data from the first and second test impedance vectors may be compared to the function. Referring now to FIG. 5A , data from the first and second test impedance vectors may include, during the first time window, RA-Can impedance measurements at 40 ohms, RV-Can impedance measurements at 36 ohms. During the second time window, the RA-Can impedance measurement may be 40 ohms and the RV-Can impedance measurement may be 33 ohms. Data from the first and second test impedance vectors can be compared to data clusters 541,542.

At 991, a body position status can be provided. In an embodiment, a comparison of chest impedance test data and a postural discriminant metric may be used to provide a postural status. In an embodiment, the body position discrimination metric may include a threshold. Comparison of the test impedance data to the threshold may determine the patient position, provided that appropriate quantitative attributes of the test impedance data meet or exceed the threshold.

In embodiments comprising clustering of the first and second chest impedance data, the position status may be provided by identifying one or more clusters associated with the chest impedance test data. For example, data point 443 in FIG. 4A falls within the upper cluster 441 . Where the upper cluster is associated with a lateral patient position, data point 443 may also correspond to a lateral patient position. Thus, it can be determined that the patient was in a lateral position during at least a portion of the first time window when the test impedance data was acquired.

Data point 444 in FIG. 4A is outside data clusters 441 and 442 . However, the association of data point 444 with one or more of one or more data clusters 441 and 442 may be evaluated. In an embodiment, data clusters 441 and 442 may correspond to acquired impedance data, such as using RA-Can impedance vector 331 and RV-Can impedance vector 332 at multiple instances during a first time window. Data points 444 may correspond to test chest impedance data acquired using RA-Can impedance vector 331 and RV-Can impedance vector 332 during the second time window. A first membership factor may be calculated using data points 444 and upper cluster 441 , and a second membership factor may be calculated using data points 444 and lower data cluster 442 . Using the comparison of the first and second attribute factors, the data point 444 can be associated with one of the data clusters 441,442. For example, the attribute factor may include the distance from data point 444 to the centroid of each of data clusters 441 and 442 . In the embodiment of FIG. 4A , the distance from data point 444 to the centroid of lower cluster 442 is less than the distance from data point 444 to the centroid of upper cluster 441 . Accordingly, data point 444 may be associated with lower cluster 442 . The association may indicate that the patient was in an upright position during at least a portion of the second time window. In an embodiment, an attribution factor may include a likelihood associated with a data point, such as to provide a likelihood that a data point is associated with a known body position or cluster of data.

10 generally illustrates an embodiment 1000, which may include obtaining first and second physiological data 1011 during a first time window, forming a function 1021 of the first physiological data versus second physiological data, obtaining physiological data during a second time window. Test data 1045, compare 1073 the physiological test data to the function, and provide a postural state 1093 using the comparison of the physiological test data and the function.

At 1011, first and second physiological data may be obtained during a first time window, as described above in the discussion of 810 of FIG. 8 . The first and second physiological data may be obtained using a plurality of different physiological sensors. The first time window may be any duration that allows for sufficiently accurate acquisition of the first and second physiological data using the plurality of physiological sensors. The time necessary to acquire data will vary depending on the type of sensor used.

At 1021, a function may be formed using the first and second physiological data. This function may be formed as discussed for 720 of FIG. 7 . In an embodiment, the function may indicate one or more of a baseline patient pathological state, a current patient physiological state, or it may identify one or more physiological trends using the first and second physiological data. The baseline patient physiological state may provide information that may be monitored or tracked relative to the patient's physiological signals, such as using processor circuit 108 . A baseline may be established by the manufacturer of the medical device, or may be established by a clinician, such as before, during, or after a device implantation procedure. In an embodiment, the baseline may be established as a moving average of recent patient information, such as the first and second physiological data, or a series of quantitative attributes associated with a function of the first and second physiological data. Metrics can also be established to provide an index or metric for comparison with a baseline. For example, a baseline can be used for comparison with test data to provide a patient positional status. Because the baseline can be a moving average, it can change over time. A changing baseline may be monitored, such as using the processor circuit 108 to monitor the baseline for deviations from previous or preset values, to ensure accurate assessment of physiological information.

At 1045, physiological test data may be obtained during a second time window. The physiological test data may be obtained in the same manner as the first and second physiological data, and the physiological test data may correspond to a second time window, which may be different from the time interval used to obtain the first and second physiological data. In an embodiment, the physiological test data may be obtained using the same physiological sensor as used to obtain one or more of the first or second physiological data, such as an accelerometer or an impedance sensor. The second and fourth physiological data may be obtained using the same physiological sensor, such as an implantable pressure sensor positioned in the patient's chest. In an embodiment, the second time window may be of a different duration than the first time window.

At 1073, the physiological test data can be compared to the function. The comparison of the physiological test data and the function may be performed by the processor circuit 108 or by external data receiving and processing circuits. The comparison can also be done manually, for example by a clinician.

In an embodiment, the physiological test data may include any parameter or characteristic determinable or measurable from patient physiological information, including, for example, data from the first test impedance vector and the second test impedance vector. The first physiological sensor can be identical to the first test impedance vector, and the second physiological sensor can be identical to the second test impedance vector. Where data from the first physiological sensor and the second physiological sensor are used to form the function, comprising at least two clusters of data, the data from the first and second test impedance vectors may be compared to the function. In an embodiment, the quantitative properties of the data from the first and second test impedance vectors may be compared to the quantitative properties of the function or cluster of data. Referring now to FIG. 5A , data from the first and second test impedance vectors may include, during the first time window, RA-Can impedance measurements at 40 ohms, RV-Can impedance measurements at 36 ohms. During the second time window, the RA-Can impedance measurement may be 40 ohms and the RV-Can impedance measurement may be 33 ohms. Data from the first and second test impedance vectors can be compared to data clusters 541,542.

At 1093, the postural status may be provided using a comparison of the physiological test data and the function. In embodiments comprising clusters of first and second physiological sensor data, the patient positional state, corresponding to at least a portion of the first time window, may be provided by identifying one or more clusters associated with the physiological test data. For example, data point 543 in FIG. 5A falls within upper cluster 541 . Thus, data point 543 corresponds to the same patient position associated with upper cluster 541, such as a side lying patient position.

Data point 544 in FIG. 5A does not fall within data clusters 541 and 542 . However, the association of data point 544 with one or more of data clusters 541 and 542 may be evaluated. For example, the distance between data point 544 and the closest data point associated with the cluster can be calculated. Data point 544 may be associated with a data cluster that includes the most recent data point. In this embodiment, data points 544 may be associated with lower cluster 542 and may be used to indicate patient position.

11 generally illustrates an embodiment 1100, which may include obtaining first and second physiological data 1111 during a first time window, forming a first function 1121 of the first physiological data versus second physiological data, using the first function to determine the second physiological data. A certain quantitative attribute 1131, obtaining third and fourth physiological data 1141 during a second time window, forming a second function 1151 of the third physiological data to the fourth physiological data, using the second function to determine a second quantitative attribute 1161, comparing First and second quantitative attributes 1175, and determining a body state 1195 using the comparison of the first and second quantitative attributes.

At 1111 , first and second physiological data may be obtained during a first time window, such as described above in the discussion of 810 of FIG. 8 . The first and second physiological data may be obtained using first and second physiological sensors, respectively, wherein the first and second physiological sensors are different. The first time window may be any duration sufficient to obtain accurate first and second physiological data.

At 1121, a first function of the first physiological data may be formed for the second physiological data. The first function may be formed as discussed for 720 of FIG. 7 .

At 1131, a first quantitative attribute can be determined using a first function. The first quantitative attribute may include any attribute of the first function, such as the area, volume or location of the data cluster, among other attributes. The first quantitative attribute may be from one or more data clusters formed using one or more parts of the function. For example, a function can describe two data clusters. The first quantitative attribute may include an expansion of the two data clusters.

In an embodiment, one or more of steps 1111, 1121 or 1131 may include a learning period of the method to determine the postural state. The learning period may include determining a postural discrimination comparison metric using the first and second physiological data, the first function, or the first quantitative attribute.

At 1141, third and fourth physiological data can be obtained during a second time window. The third and fourth physiological data may be obtained in the same manner as the first and second physiological data, although the third and fourth physiological data may be obtained during a different time interval, such as a second time window. In an embodiment, the first and third physiological data may be obtained using the same physiological sensor, such as an impedance sensor configured to obtain cardiac impedance vector information. The second and fourth physiological data may be obtained using the same physiological sensor, such as an implantable pressure sensor positioned in the patient's chest. In an embodiment, the second time window may be of a different duration than the first time window.

At 1151, a second function may be formed using the third physiological data and the fourth physiological data. The second function may be formed in the same manner as the first function, or as described above in the discussion of 820 of FIG. 8 .

At 1161, a second quantitative attribute can be determined using a second function. The second quantitative property may include any property of the second function, such as including the same properties described above in the discussion of 830 of FIG. 8 .

In an embodiment, one or more of steps 1141 , 1151 or 1161 may include a test phase of the method to determine the postural state. The testing session may include obtaining information for comparison with a postural discriminant comparison metric, such as using third and fourth physiological data, a second function, or a second quantitative attribute.

At 1175, the first and second quantitative attributes can be compared. In an embodiment, the first and second quantitative attributes may include impedance magnitude information, centroid location information, or volume information. In an embodiment, the first and second areas may be compared. In an embodiment, two or more quantitative attributes may be compared, such as for trending. For example, first, second, and third centroid locations may be evaluated, e.g., in association with first, second, and third time windows, to determine trends in centroid locations. Correlation of the fourth centroid position, as corresponding to the test chest impedance data, with the trend can be evaluated.

In an embodiment, time intervals associated with quantitative attributes may be used to form a lookup table. Physiological test data can be obtained and compared to a lookup table, eg, to provide postural status. In an embodiment, the first, second, and third centroid locations discussed above may each be associated with a discrete patient position, or discrete portion of a look-up table. The best fit of the fourth centroid location within the lookup table can be evaluated.

At 1195, the postural state can be determined using the comparison of the first and second quantitative attributes. In an embodiment, the area of the data cluster associated with the first time window may be compared to the area of the data cluster associated with the second time window. Patient position can be applied to the data points included in the overlapping region.

additional notes

Embodiment 1 includes subject matter, such as a medical device, comprising: a processor comprising a first data input configured to receive first physiological data from a first physiological sensor, the first physiological data corresponding to a plurality of instances during, and a second data input configured to receive second physiological data from a second physiological sensor, the second physiological data corresponding to the plurality of instances during the same first time window. Embodiment 1 may include subject matter such as a first data input configured to use a first physiological sensor to obtain first physiological data, and a second data input configured to use a second physiological sensor to obtain second physiological data. Embodiment 1 may include subject matter, such as a processor-readable medium, comprising instructions that, when executed by a processor, configure a medical device to form a function of first physiological data on second physiological data, using the function to form at least two data clusters, determine quantitative properties of at least one of the data clusters, and use the quantitative properties to provide an index of heart failure decompensation.

In Example 2, the subject matter of Example 1 may optionally include a first data input configured to receive first physiological data comprising a first electrode configuration that defines a first thoracic impedance vector. first thoracic impedance data, and a second data input configured to receive second physiological data comprising a second thoracic impedance obtained using a different second electrode configuration defining a different second thoracic impedance vector Impedance data.

In Example 3, the subject matter of one or any combination of Examples 1-2 can optionally include a first data input configured to receive first physiological data comprising using a location located at First chest impedance data obtained from a first electrode and at least a different second electrode in or near a ventricle of the heart.

In Example 4, the subject matter of one or any combination of Examples 1-3 can optionally include a first data input or a second data input configured to be coupled to an accelerometer to obtain the first physiological data or the second physiological data. At least one of the data inputs.

In Example 5, the subject matter of one or any combination of Examples 1-4 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to use body position The information forms at least two data clusters.

In Example 6, the subject matter of one or any combination of Examples 1-5 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to use body position information to determine the first and second physiological data for forming the function.

In Example 7, the subject matter of one or any combination of Examples 1-6 can optionally include a first data input configured to receive first physiological data from a first physiological sensor, the first physiological The data corresponds to a plurality of instances during the first time window including at least one change in patient position.

In Example 8, the subject matter of one or any combination of Examples 1-7 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to use the time information to determine the first and second physiological data for forming the function.

In Example 9, the subject matter of one or any combination of Examples 1-8 can optionally include a first data input configured to receive first physiological data from a first physiological sensor, the first data input Corresponding to a plurality of instances during a first time window comprising time intervals in which a change in patient position is expected to occur.

In Example 10, the subject matter of one or any combination of Examples 1-9 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to determine data A quantitative property of at least one of the clusters, the quantitative property comprising at least one of the following: an expansion of a portion of a function, an extent of a portion of a function, an area at least partially bounded by a portion of a function, a location or shape of a portion of a function The centroid, the position of at least two different parts of the function or the distance between centroids, or a different second function formed using at least two different parts of the function.

In Example 11, the subject matter of one or any combination of Examples 1-10 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to compare a baseline Distance is the distance between the positions or centroids of at least two distinct parts of the function.

In Example 12, the subject matter of one or any combination of Examples 1-11 can optionally include a third data input configured to receive additional physiological data from at least a third physiological sensor, the additional physiological data corresponding to multiple instances during the same first window of time, and a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to use the first physiological data, the second physiological data, and the Additional physiological data form a multidimensional function.

In Example 13, the subject matter of one or any combination of Examples 1-12 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to determine data A quantitative property of at least one of the clusters includes determining volume using a portion of the multidimensional function.

In Example 14, the subject matter of one or any combination of Examples 1-13 can optionally include a first data input configured to receive third physiological data from the first physiological sensor, the third physiological data Corresponding to a plurality of instances during a second time window, a second data input configured to receive fourth physiological data from a second physiological sensor corresponding to a plurality of instances during the same second time window Example, and a processor-readable medium comprising instructions that, when executed by a processor, configure a medical device to form a second function of third physiological data to fourth physiological data, using the second function to form at least two Additional data clusters, determining a test quantitative attribute of at least one of the additional data clusters, and using the test quantitative attribute to provide an index of heart failure decompensation.

In Example 15, the subject matter of one or any combination of Examples 1-14 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to quantify Attributes and Tests Quantitative attributes were trended to provide indicators of heart failure decompensation.

In Example 16, the subject matter of one or any combination of Examples 1-15 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure the medical device to use the time The information forms at least two data clusters.

Example 17 may comprise or may combine the subject matter of one or any combination of Examples 1-16 to optionally comprise subject matter, such as a system comprising an implantable medical device comprising a first physiological sensor which configured to obtain first physiological data corresponding to a plurality of instances during a first time window, a second physiological sensor configured to obtain second physiological data corresponding to a plurality of instances during the same first time window, and a processor circuit configured to: receive first and second physiological data, use the first and second physiological data to form at least two different data clusters, and use a quantitative attribute of at least one of the data clusters to provide heart failure Indicator of decompensation.

In Example 18, the subject matter of one or any combination of Examples 1-17 can optionally include an impedance measurement circuit configured to receive a first impedance signal using at least a first physiological sensor and to use a second physiological sensor A second impedance signal is received, wherein the first physiological sensor includes a first electrode and the second physiological sensor includes a second, different electrode.

In Example 19, the subject matter of one or any combination of Examples 1-18 can optionally include a storage circuit configured to store a plurality of quantitative attributes.

Embodiment 20 may comprise or may combine the subject matter of one or a combination of Embodiments 1-19 to optionally comprise subject matter such as a medical device comprising: a processor comprising a first data input configured to receive data from a second first chest impedance data for a chest impedance vector, the first chest impedance data corresponding to a plurality of instances during the first time window, a second data input configured to receive a second chest impedance data from a second chest impedance vector Impedance data, the second chest impedance data corresponding to multiple instances during the same first time window, and a third data input configured to receive test chest impedance data using at least two chest impedance vectors.

In Example 21, the subject matter of one or any combination of Examples 1-20 can optionally include a processor-readable medium comprising instructions that, when executed by a processor, configure a medical device to: form a first a first function of chest impedance data on a second chest impedance data, using the first function to form a first pair of data clusters, using the first pair of data clusters to determine a first quantitative attribute, using the test chest impedance data to form a second function, using The second function forms a second pair of data clusters, forms a second quantitative attribute using the second pair of data clusters, and provides a heart failure decompensation index using a comparison of the first and second quantitative attributes.

These non-limiting examples can be combined in any permutation or combination.

The above detailed description includes references to the accompanying drawings, which form a part of the detailed description. The drawings show, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are also referred to herein as "examples." Such embodiments may include elements in addition to those shown or described. However, the inventors also contemplate embodiments in which only those elements shown or described are provided. In addition, the inventors contemplate using those elements shown or described (or any combination or permutation of one or more aspects thereof).

In the event of inconsistent usage between this document and those documents incorporated by reference, the usage in this document controls.

In this document, the term "a" or "an" is used, as is common in patent literature, to include one or more than one, independently of "at least one" or "one or more any other instance or use of ". In this document, the term "or" is used to mean a non-exclusive, alternative, such that "A or B" includes "A but not B," "B but not A" and "A and B," unless otherwise illustrate. In this document, the terms "including" and "in which" are used as the plain English equivalents of the corresponding terms "comprising" and "wherein". Likewise, in the following claims, the terms "comprising" and "comprising" are open ended, that is, a system, device, article, or system that includes elements other than those listed after such term in a claim. The process is still considered to be within the scope of the claim. Furthermore, in the following claims, the terms "first", "second", and "third", etc. are used only as labels and are not intended to impose numerical requirements on their objects.

The method embodiments described herein may be implemented by a machine, a processor, or at least partially a computer. Some embodiments may include a computer-readable medium or a processor-readable medium encoded with instructions operable to configure an electronic device to perform the methods described in the above embodiments. The implementation of this method can include code such as microcode, assembly language code, higher level language code, etc. Such code may include computer readable instructions for performing various methods. The code may form part of a computer program product. Additionally, in an embodiment, the code may be tangibly stored on one or more volatile, non-transitory, or non-volatile tangible computer-readable media, such as during execution, or at other times. Examples of such tangible computer-readable media may include, but are not limited to, hard disks, removable disks, removable optical disks (e.g., compact disks, digital video disks), magnetic tape cartridges, memory cards or sticks, random access memory (RAM), Read memory (ROM), etc.

The above description is intended to be illustrative, not restrictive. For example, the above-described embodiments (or one or more aspects thereof) may be used in combination with each other. Other implementations may be used, as would be used by one of ordinary skill in the art after reviewing the above description. The Abstract is provided to comply with 37 C.F.R. §1.72(b) to allow the reader to quickly ascertain the nature of the technical disclosure. It is to be understood that it will not be used to interpret or limit the scope or meaning of the claims. Also, in the above Detailed Description, various features may be grouped together to simplify the disclosure. This should not be interpreted as intending that an unclaimed disclosed feature is essential to any claim. Rather, inventive subject matter may lie in less than all features of a particular disclosed embodiment. Thus the following claims are hereby incorporated into the Detailed Description, with each claim standing on its own as a separate embodiment, and it is contemplated that such embodiments may be combined with each other in various combinations and permutations. The scope of the invention should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims (15)

1. A medical device for providing an indicator of heart failure decompensation, said medical device comprising: a processor, the processor comprising: means for obtaining a first data input configured to receive first time-varying physiological data from a first physiological sensor, the first time-varying physiological data corresponding to multiple instances during the window; and means for obtaining a second data input configured to receive second time-varying physiological data from a second physiological sensor, the second physiological data corresponding to during the same first time window Multiple instances of ; and When the processor executes the instructions in the non-transitory processor-readable medium, the medical device is configured to: forming a function of the first physiological data to the second physiological data, the function having a multivalued output; identifying at least one boundary in the multivalued output of the function to separate the multivalued output into respective data clusters; determining a quantitative attribute of at least one of the data clusters; and The quantitative attributes are used to provide an index of heart failure decompensation. 2. The medical device of claim 1, wherein said first data input is configured to receive as said first physiological data first chest impedance data obtained using a first electrode configuration defining a first chest impedance vector; and Wherein the second data input is configured to receive as the second physiological data second chest impedance data obtained using a different second electrode configuration defining a different second chest impedance vector. 3. The medical device of claim 2, wherein said first data input is configured to receive said first thoracic impedance using a first electrode positioned in or near a ventricle of the heart, and at least a different second electrode data. 4. The medical device of claim 1, wherein at least one of the first data input or the second data input is configured to be coupled to an accelerometer to obtain the first physiological data or the second physiological data. 5. The medical device of claim 1 , wherein when the processor executes the instructions in the non-transitory processor-readable medium, the medical device is further configured to use body position information to identify one of the multivalued outputs of the function At least one boundary. 6. The medical device of claim 1 , wherein when the processor executes the instructions in the non-transitory processor readable medium, the medical device is further configured to use body position information to determine the first and second physiological data to use to form the function. 7. The medical device of claim 6, wherein said first data input is configured to receive said first physiological data from said first physiological sensor, said first physiological data corresponding to Multiple instances of , including at least one patient position change. 8. The medical device of claim 1, wherein when the processor executes the instructions in the non-transitory processor readable medium, the medical device is further configured to use time-of-day information to determine the first and second physiological data to use to form the function. 9. The medical device according to claim 8, wherein said first data input is configured to receive said first physiological data from said first physiological sensor, said first physiological data corresponding to During multiple instances, the first time window includes a time interval in which a change in patient position is expected to occur. 10. The medical device of claim 1 , wherein when the processor executes the instructions in the non-transitory processor-readable medium, the medical device is further configured to determine a quantitative attribute of at least one of the data clusters, The quantitative attributes include at least one of the following: expansion of parts of said function; the scope of the portion of the function; an area defined at least in part by a portion of said function; the location or centroid of the portion of the function; the distance between the positions or centroids of at least two different parts of said function; or A different second function is formed using at least two different parts of the function. 11. The medical device of claim 10 , wherein when the processor executes the instructions in the non-transitory processor-readable medium, the medical device is further configured to compare the baseline distance with at least two different parts of the function The locations or distances between centroids are compared. 12. The medical device of claim 1 , wherein when the processor executes the instructions in the non-transitory processor readable medium, the medical device is further configured to obtain a third data input, the third data input being configured to receive additional physiological data from at least a third physiological sensor, the additional physiological data corresponding to a plurality of instances during the first time window; and Wherein when the processor executes the instructions in the non-transitory processor-readable medium, the medical device is further configured to use the first physiological data, the second physiological data and the additional physiological data to form a multidimensional function. 13. The medical device of claim 12, wherein when the processor executes the instructions in the non-transitory processor-readable medium, the medical device is further configured to determine a quantitative attribute of at least one of the data clusters, Partially determined volumes are included using the multidimensional function. 14. The medical device of claim 1, wherein the first data input is configured to receive third physiological data from the first physiological sensor, the third physiological data corresponding to a plurality of instances during the second time window ;and wherein the second data input is configured to receive fourth physiological data from the second physiological sensor, the fourth physiological data corresponding to a plurality of instances during the second time window; and Wherein when the processor executes the instructions in the non-transitory processor-readable medium, the medical device is further configured to: forming a second function of said third physiological data on said fourth physiological data; identifying at least one boundary in the multivalued output of the second function to separate the multivalued output into respective additional data clusters; determining a test quantitative attribute of at least one of the additional data clusters; and The quantitative properties of the test are used to provide the heart failure decompensation index. 15. The medical device of claim 14 , wherein when the processor executes the instructions in the non-transitory processor readable medium, the medical device is further configured to trend the quantitative attribute and the test quantitative attribute To provide the heart failure decompensation index.