Every story in this issue with a short summary, so you can decide what to read before leaving the site. Summaries are written by AI (Claude) only from each publisher's own description and checked by Jev for anything not in that description. They are not full translations. Follow "Read the original" for the full story.
A case-level audit of a public thyroid imaging archive found contradictory labels and data splits prone to leakage. It also found strong evidence that deep learning models exploit acquisition artifacts.
Read the original →This is the seventh installment in a series on medical imaging AI infrastructure. Earlier installments covered HIRA's first AI-assisted claims review, platform competition, lung cancer screening AI, breast AI, stroke AI and DICOM standardization.
Read the original →A University of Tokyo feasibility study finds that AI-based throat imaging for influenza could overcome the biological sampling bottleneck that has limited fully remote diagnosis. This holds provided four conditions are met.
Read the original →Large AI models, such as large language models, have shown expert-level performance in specific medical domains. Concerns about medical bias are growing, but related research is fragmented and shaped by discipline-specific assumptions, terminology and evaluation criteria. A general understanding of medical bias in large AI models is still lacking.
Read the original →Artificial intelligence is reshaping the healthcare sector by improving the way medical professionals work.
Read the original →A machine learning model was trained on more than 12,000 infant screenings. It uses twelve clinical and demographic variables to identify which babies need hip ultrasound, and it achieved strong external validation.
Read the original →Children’s National Hospital in D.C. says its goals are to shorten wait times and make sure families don’t leave before they’re seen.
Read the original →Executives from Blitz and Shift Technology at ITC Vegas weigh the cost of AI as token prices shift.
Read the original →AI has changed the pace of biological discovery. The next test for industrial biotechnology is whether manufacturing can keep pace.
Read the original →A group of scientists led by Joseph Brown, an assistant professor in the University of Toronto's Leslie Dan Faculty of Pharmacy, is developing an open-access database and an artificial intelligence training model. The tools are intended to identify peptides with therapeutic potential.
Read the original →A new study finds that large language models rewriting medical literature for patients retain only about half of the original knowledge. Omissions persist even after corrective prompting.
Read the original →Deep Health AI India Ltd submitted a revised audit report for FY26 to include entity names in the consolidated financial statements, confirming no change in financial results. The auditor flagged key matters regarding an exceptional loss on a proposed acquisition and the deviation in utilization of Rights Issue proceeds.
Read the original →Medical artificial intelligence startup OpenEvidence and competitors Doximity Inc. and Pathway Medical Inc. told a Massachusetts federal judge on Thursday that they have reached a settlement in principle. The deal would resolve a contentious lawsuit and counterclaim.
Read the original →The European Publishers Council (EPC) is highly concerned about an update to Meta's new commercial terms. The update will allow the company to scrape content from publishers.
Read the original →Researchers at the Else Kroner Fresenius Center (EKFZ) for Digital Health at TU Dresden and Dresden University Hospital have developed an on-premises medical AI system.
Read the original →Esophageal cancer is often diagnosed at an advanced stage, so early detection is important for improving prognosis. Current screening methods such as upper endoscopy are invasive and limited by medical resources, making large-scale screening of asymptomatic populations difficult. A Chinese research team therefore studied an approach that does not use endoscopy.
Read the original →