The NIH has eliminated its MOSAIC transition-grant program and sharply reduced other early-career awards, leaving more than a hundred young scientists without the funding they were promised to start their own labs. Officials say they are removing programs tied to diversity initiatives, but researchers warn the cuts are undermining scientific careers and disproportionately harming scientists from disadvantaged backgrounds who already face barriers in academia.

Advance patient safety, quality, and leadership with Harvard Medical School's Safety, Quality, Informatics, and Leadership certificate program. Gain tools to improve outcomes, drive system-wide change, and lead data-driven health care improvement.

HL7's Vulcan FHIR Accelerator is advancing the adoption of FHIR standards in clinical research through its Vulcan Interoperability Bridge. Program manager Sandy Vance highlights the initiative's success in streamlining processes, such as adverse event reporting, by connecting stakeholders, including the FDA and EHR vendors. The VIB aims to enhance operational efficiencies and data accessibility, with plans to expand into Europe and Japan. The initiative's first cohort focused on areas like internal sponsor systems and patient engagement, with a second cohort set to further integrate FHIR-to-OMOP data transformations.

AI can enhance epilepsy care by identifying patients who might benefit from surgery and spotting missing evaluations, according to a study presented at the American Epilepsy Society meeting. The research found that AI could streamline treatment by filling gaps in patient records, such as outdated MRIs and missing neuropsychological assessments. This approach aims to reduce treatment delays and improve patient outcomes.

Gena Lawday, chief quality officer at UVA Community Health, warns of the risks associated with AI-driven clinical decision support tools, particularly algorithmic drift, which can lead to a gradual decline in accuracy as patient populations and care practices evolve. This concern arises as hospitals increasingly adopt AI for tasks like sepsis alerts and risk scoring. Lawday emphasizes the need for continuous monitoring of AI tools to ensure they remain reliable and safe. "The dangerous part of that is the degradation can be subtle," Lawday said. "Clincians won't necessarily see a clear failure, just a gradual decline in reliability that can lead to missed deterioration or unnecessary workups."

At Ochsner Health, AI is transitioning from experimental projects to a core clinical capability, with Innovation Officer Jason Hill emphasizing its potential to enhance care quality and efficiency. Hill highlights the importance of integrating AI into health care workflows, focusing on tools like OpenEvidence and ambient documentation systems that reduce clerical burdens. He stresses the need for continuous monitoring of AI models to manage drift and ensure safety. Hill advises health system CIOs to prioritize AI investments in lower-risk operational areas and to leverage existing productivity platforms.

Athenahealth is set to integrate Microsoft Dragon Copilot into its EHR system, athenaOne, by mid-2026. This collaboration will enhance Athenahealth's Ambient Notes, allowing clinicians to choose their preferred ambient documentation model. The integration aims to streamline clinical note-taking, enabling health care providers to focus more on patient care. Currently, over 170,000 clinicians across nearly 50 specialties use Ambient Notes, and the addition of Dragon Copilot will expand their capabilities with AI-generated clinical notes.

Scientists are making strides in using AI to design antibodies, potentially transforming therapeutic development. Recent advancements have led to the creation of nanobodies and full-length antibodies with properties similar to commercial drugs. Key players in this field include Nabla Bio, Chai Discovery and the University of Washington. While initial AI-designed antibodies lacked potency, new tools have improved their design, targeting diseases like cancer and infections. These developments could accelerate the introduction of AI-designed therapeutics into clinical trials.

Google has fixed a critical flaw in Gemini Enterprise that allowed threat actors to steal sensitive corporate data by embedding malicious instructions in shared documents without user interaction. The vulnerability, discovered by Noma Labs and named GeminiJack, exploited Gemini Enterprise's access to organizational Gmail, Docs and Calendar.

Researchers at the University of Exeter have developed a genetic risk score to predict the likelihood of developing type 1 diabetes, which could be used in large-scale health studies. This test identifies individuals at high risk, allowing for early intervention with drugs like teplizumab, which can delay the onset of the disease. The paper, published in The Lancet Diabetes & Endocrinology, highlights the importance of early detection, as the drug is only effective before symptoms appear.

Six major health organizations including the American Academy of Pediatrics and the American College of Physicians have opposed the Advisory Committee on Infection Practice's recommendation to limit the hepatitis B vaccine at birth. Representatives from the groups argue that postponing newborn vaccination could result in preventable child deaths and undermine public health gains, citing research showing the change could result in hundreds of additional cases of liver cancer, 480 preventable deaths and significant yearly increases in health care costs.