What We’re Reading

A new spin on an old section, “What We’re Reading,” is now a peek inside the minds of the NIH scientific community. Here, we highlight commentaries recently published that challenge the scientific status quo.

NIA: The interneuron hypothesis of amyotrophic lateral sclerosis, Brain

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disorder that affects the motor neurons—so, unsurprisingly, much of the research in the field has focused on these neurons. However, motor neurons might not tell the whole story. Interneurons, which connect sensory and motor neurons, have been shown to be damaged in people with ALS, particularly in their primary motor cortexes. A study of the genomes of 20,000 ALS patients linked GABAergic cortical interneurons to ALS pathogenesis. Neurophysiological data also show that impaired functioning of inhibitory cortical interneurons may be responsible for cortical hyperexcitability—a common feature of ALS. It’s not entirely clear why these interneurons are being damaged in people with ALS. One theory is that the overfiring of excitatory neurons has the downstream effect of exhausting the inhibitory capabilities of the interneurons. Another potential explanation is that there’s a genetic component to more directly leading to interneuron death. Mechanisms aside, it’s become clear that interneurons play a significant role in ALS pathogenesis. (NIH author: B.J. Traynor, PMID: 40179249)

[BY AMELIA MARVIT, NIAID]

OIR: Equipping AI for Unbiased and Inclusive Neurology, JAMA Neurology

As artificial intelligence (AI) becomes increasingly integrated into medicine, it is essential that the data used to train these systems reflect the full diversity of the population. To advance personalized medicine and achieve truly individualized outcomes, AI must include outlier populations such as children, elderly adults, racial and ethnic minorities, and individuals facing economic or social disadvantages. Neurologists and other clinicians are essential partners in this effort. Without their direct involvement in developing and validating AI tools and large language models, there is a real risk that these underrepresented groups will be excluded from the knowledge base these technologies rely on. This can lead to care shaped by technologies that fail to recognize their own limitations and may generate inaccurate or biased outputs to fill informational gaps. By involving medical experts and ensuring inclusive, representative data, AI can move toward more equitable, informed, and effective health care for all. (NIH author: N.F. Schor, PMID: 39585712)

[BY KAMRYN CREGGER, NIAID]

ONR: This is the moment: advancing “Food is Medicine” through research, American Journal of Clinical Nutrition

Food insecurity is a driver of obesity, diabetes, and cardiovascular disease. As a cornerstone of health, nutrition plays a critical role in preventing disease and promoting wellness. ONR is addressing these challenges through coordinated, cross-agency efforts.

For example, one project is evaluating the impact of medically tailored food pantries for patients with cancer; others look at prescriptions for produce and videoconference-based cooking classes. Many of these projects assess cost-effectiveness and the majority are randomized clinical trials. The NIH, in partnership with the All of Us Research Program, is also supporting the largest national investment in nutrition research, Nutrition for Precision Health.

Given the complex factors influencing nutrition-related conditions, a multidisciplinary, cross-sectoral research approach is vital—an effort in which the NIH plays a central role. The goal is to generate publicly accessible data that can inform targeted, affordable, and accessible nutrition strategies to prevent disease, reduce metabolic risk, improve overall health, and enhance quality of life at both individual and population levels. (NIH authors: N.J. Jury, E.G. Guillen, and A.A. Bremer, PMID: 39900117)

[BY KAMRYN CREGGER AND AMELIA MARVIT, NIAID]

NCI-CCR: Hallmarks of artificial intelligence contributions to precision oncology, Nature Cancer

Cancer’s inherent heterogeneity poses a major challenge in oncology, as no single therapy works for all patients. Precision oncology addresses this problem by tailoring treatments to the unique molecular and genetic features of individual tumors. Artificial intelligence (AI) is accelerating this shift, transforming key areas of cancer care—including early detection, diagnosis, treatment planning, drug discovery, and clinical trial design. AI’s influence spans ten core domains: Screening, detection, tumor profiling, outcome prediction, treatment monitoring, clinical trial optimization, biomarker discovery, therapeutic combinations, vulnerability identification, and drug design. Notably, AI has achieved regulatory approvals and diagnostic accuracy on par with, or exceeding, human experts, while also expediting drug development. By integrating genomic, imaging, and clinical data, AI enables more personalized, predictive, and adaptive care. Despite this progress, challenges remain. Issues related to data quality, model generalizability, and the need for rigorous validation limit widespread adoption. To fully realize AI’s potential, interdisciplinary collaboration is essential—engaging researchers, clinicians, patients, industry, payers, and regulators. The NCI-CCR is actively addressing these challenges through translational research aimed at real-world application. With sustained investment and a focus on transparency, interpretability, and equitable access, AI is poised to reshape oncology into a more precise, effective, and patient-centered field, ultimately helping to reduce the global burden of cancer. (NIH authors: T.-G. Chang, S. Park, A.A. Schäffer, P. Jiang, and E. Ruppin, PMID: 40055572)

[BY KAMRYN CREGGER, NIAID]