Volume 31 Issue 10, October 2025

A trio of treatments could block the inheritance of some severe diseases, but more than science must advance.

When clinical trials are suddenly halted by US funding cuts, there are repercussions for investigators and patients, and they do not stop at borders.

Ophthalmology could be the first medical specialty fundamentally transformed by artificial intelligence. Pearse Keane recalls an early breakthrough in this journey.

Genomic sequencing promises targeted treatment for every patient with cancer; the reality is much tougher to deliver.

Obesity and associated complications can be managed by obesity medications, prompting the revision of criteria for the diagnosis and staging of this disease.

DeepSeek’s potential for on-premises deployment in hospitals reveals a regulatory ‘gray area’ that requires enhanced guidance.

To facilitate the safe and effective implementation of autonomous artificial intelligence agents in healthcare, regulatory frameworks must evolve beyond static device paradigms to incorporate adaptive oversight and flexible pathways.

The US Food and Drug Administration (FDA) is at a crossroads. There is an urgent need to strengthen the integrity of drug approvals and rejections by diffusing the decision-making authority of individual officials and improving transparency when disagreements arise.

Germline sequencing should be integrated with somatic testing of cancers to provide a comprehensive genetic assessment of disease.

A clinical trial designed to overcome limitations of previous trials now confirms that individuals consume more calories from ultra-processed diets than from minimally processed diets, even when both diets meet UK dietary guidelines and participants are losing weight.

Despite mixed efficacy outcomes, a phase 1/2 umbrella trial brings the glioblastoma field closer to a new era of diagnostic precision and targeted therapy.

A new study provides physiological insights into the dying process, which could have substantial implications for resuscitation science, death determination and organ donation.

Imaging data support the clinical effects of RNA interference agent vutrisiran in patients with transthyretin amyloidosis with cardiomyopathy — and point to a potential new role for echocardiographic parameters in monitoring individual treatment responses.

We studied people with obesity involved in the SURMOUNT-1 trial of the dual GLP-1 and GIP receptor agonist, tirzepatide. People with pathogenic variants of the gene encoding the melanocortin 4 receptor (MC4R) lost a similar amount of weight to people with an intact gene, demonstrating that tirzepatide is effective in MC4R deficiency.

Malaria parasites that evade the most widely used rapid diagnostic tests are spreading. We combined new data and modeling to pinpoint global hotspots at risk for these diagnostic-resistant parasites and to guide surveillance and contingency plans.

We created and curated three benchmark datasets to assess large language model (LLM) performance on sleep and fitness tasks ranging from answering expert questions to real-world coaching scenarios. Fine-tuning the Gemini LLM on real-world coaching tasks and self-reported sleep-quality outcomes improved its performance and provided a benchmark for further development.

We introduce a diagnostic biomarker for amyotrophic lateral sclerosis that distinguishes it from similar conditions by examining a combination of plasma proteins. This innovation also offers insights into the underlying mechanisms of the disease, highlighting a compensatory process in skeletal muscle and bioenergetics that seems to start much earlier than previously thought.

We assessed the utility of RNA sequencing in molecular diagnostics by analyzing a large pan-cancer cohort ranging in age from infants to seniors. This technique is highly sensitive for the identification of gene alterations that can guide diagnosis and targeted therapy, and enables additional biological and clinical interpretation of findings beyond standard DNA-sequencing protocols.

This Review summarizes recent technical advancements in generative AI, outlines how new models might improve healthcare and discusses validation approaches—using lessons from recent successes and failures in the field.

The STARD-AI statement, which was developed through a multistage process involving over 240 international stakeholders, provides a minimum set of criteria that allows for comprehensive reporting of AI-centered diagnostic test accuracy studies.

Analyzing data from the SURMOUNT-1 trial, the authors show that the weight loss effects of tirzepatide treatment are similar in carriers and noncarriers of MC4R pathogenic mutations.

A free-living trial in people with overweight or obesity found that minimally processed diets led to greater weight loss and cardiometabolic improvements than ultraprocessed diets following UK healthy eating guidelines at 8 weeks.

A clinical trial using continuous glucose monitoring found that a Dietary Approaches to Stop Hypertension (DASH)-style diet tailored for diabetes reduced hyperglycemia and improved glucose control compared to a typical American diet in patients with type 2 diabetes.

A systematic review and meta-analysis of clinical trials investigating the efficacy and safety of obesity management medications reported tirzepatide and semaglutide to be the most effective in reducing body weight and obesity-related complications.

A post hoc analysis of a multicentre, randomised trial showed that prediabetes remission is possible without total weight loss—providing weight is distributed to subcutaneous deposits as opposed to visceral ones.

In an observational study comparing individuals with type 2 diabetes and obesity who underwent metabolic surgery with similar individuals who received GLP-1 receptor agonist treatment, metabolic surgery was associated with a lower risk of macrovascular and microvascular outcomes, including major adverse cardiovascular events, nephropathy, retinopathy and all-cause mortality.

An integrated dataset combining genetics, epigenomics, transcriptomics, proteomics and metabolomics from 1,342 people living with HIV illuminates molecular pathways driving immune responses and comorbidities in this population.

Serological profiles from two observational cohort studies in The Gambia characterize surrogates of protection against Streptococcus pyogenes and reveal an antibody response against antigens that are currently in vaccine development.

A modeling approach to assess the global risk of Plasmodium falciparum histidine-rich protein 2 and 3 gene deletions identifies ten priority African countries for surveillance and could inform malaria control policies.

High-resolution geospatial mapping found that the annual incidence of cholera shifted from western to central and eastern Africa between 2011 and 2020, with the latter regions more likely to report cholera in 2022–2023, reflecting instability in cholera burden patterns that can impact progress in disease control.

In this attempt at xenotransplantation of a lung from a genetically modified pig into a brain-dead recipient, although the grafted lung initially maintained viability and functionality, antibody-mediated rejection rapidly occurred, contributing to xenograft damage.

A large language model designed for health monitoring provides personalized sleep and fitness predictions, insights and advice.

Trained and validated on multimodal data from 14.5 million images from multicountry datasets, a foundation model is shown to increase diagnostic and referral accuracy of clinicians when used as an assistant in a trial involving 16 ophthalmologists and 668 patients.

Using large cohorts from published clinical trials involving more than 8,000 patients with multiple sclerosis, a probabilistic machine learning model reconstructs the transition probabilities from data-derived diseases statuses, showing patterns that suggest how progression to severe stages occur and potential inversion of the process.

A randomized controlled trial found no difference in paranoid ideations between virtual reality-based and gold-standard cognitive behavioral therapy for patients with schizophrenia spectrum disorders, in contrast to previous trials using wait-list or passive controls.

Plasma proteomic analysis of presymptomatic individuals with amyotrophic lateral sclerosis identifies novel disease-associated proteins and enables the development of an accurate diagnostic biomarker panel.

This pilot trial showed that perioperative treatment with the isocitrate dehydrogenase (IDH) inhibitor safusidenib of patients with low-grade IDH-mutant glioma, with craniotomy and lumbar puncture before and after treatment, is feasible and safe and enabled in-depth translational investigation of safusidenib treatment-induced changes in the tumor, including electrophysiological effects.

A combination of mouse model data and data from a randomized phase 2 trial in patients with breast cancer in remission demonstrates feasibility, safety and a reduction in residual tumor cells following treatment with hydroxychloroquine and/or everolimus.

Five-year follow-up and prespecified exploratory biomarker analysis from the phase 3 KEYNOTE-426 trial show that pembrolizumab plus axitinib compared to sunitinib as first-line therapy continues to show overall and progression-free survival benefits in patients with advanced clear cell renal cell carcinoma and indicate that a T-cell-inflamed gene expression profile is predictive of response.

In the dose-expansion cohort of this phase 1b trial, the EGFR×HER3 bispecific antibody–drug conjugate BL-B01D1 given to patients with metastatic esophageal squamous cell carcinoma led to encouraging clinical response rates, which need to be confirmed in larger efficacy trials.

In this single-arm phase 2 trial in patients with HR⁺HER2⁻ advanced breast cancer, treatment with the HER3-targeting antibody–drug conjugate paritumab deruxtecan led to encouraging objective response rates, and comprehensive exploratory analyses indicate potential biomarkers of response.

On the basis of encouraging preclinical activity, an anti-CEACAM5 antibody–drug conjugate with a topoisomerase 1 inhibitor payload is being tested in an ongoing phase 1 trial where recommended doses for further testing have been identified and an unconfirmed overall response rate of 15% was observed in patients with metastatic colorectal cancer.

In the proof-of-concept phase 2 ROME trial, comprehensive genomic profiling followed by molecular tumor board evaluation and randomization of patients with metastatic solid cancer to receive personalized therapy or standard of care led to a significantly higher objective response rate and longer progression-free survival in patients who received personalized therapy.

In a prospective study including 2,300 pediatric and adult patients with liquid and solid tumors, RNA sequencing provided actionable molecular information in 87% of cases, with applications in diagnostics and therapy matching.

In a phase 1/2a umbrella trial in patients with glioblastoma matching targeted therapies according to tumor molecular profiling, temsirolimus plus radiotherapy improved PFS compared to standard of care in patients with phospho-mTOR activation.

The authors conducted prospective multimodal monitoring of simultaneous brain and heart function to define physiological changes during the human dying process leading to circulatory arrest.

In a randomized trial of patients with a previous myocardial infarction and with residual inflammation, a monoclonal antibody targeting the LOX-1 receptor, thought to contribute to atherosclerotic plaque progression and inflammation, reduced soluble LOX-1 and interleukin-6 levels but did not lead to a substantial reduction in coronary plaque volume.

In the HELIOS-B randomized trial in patients with transthyretin amyloid cardiomyopathy, treatment with the RNA interference agent vutrisiran, which targets transthyretin, led to improved echocardiographic measures of cardiac structure and function, as compared to placebo.