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 new study finds that AI software can extract reliable bone mineral density measurements from routine non-contrast CT scans. The measurements match DXA well enough to enable opportunistic osteoporosis screening.
Read the original →A machine learning model called Amylo-Detect uses routine electronic health record data to identify patients at risk for cardiac amyloidosis. It outperformed existing scores and caught cases that were missed.
Read the original →Implementation costs, structural pressures and the labor-intensive nature of healthcare mean that artificial intelligence is unlikely to dramatically lower costs. This is expected to hold at least in the short term.
Read the original →Epic Systems used Anthropic's Claude Mythos AI model to identify security vulnerabilities that could expose patient health records to undetected access. The company paused parts of product development to implement patches addressing the risks.
Read the original →LT350's first pilot canopy is advancing at a Dallas-area hospital. Auddia's letter cites AI-ready construction costs of $15–$35 million per megawatt.
Read the original →Interest is rising in companies handling DNA synthesis, protein expression and experimental validation as AI drug discovery advances. AI can now design protein, antibody and small-molecule candidates quickly, but experiments are still needed to confirm they work. In an example cited by Ultimator Capital partner Freda Duan, the AI model Claude designed 1,320 protein binders, and 354 were confirmed to bind to the target.
Read the original →The agency will award millions of dollars in prizes to automate clinical evidence synthesis and develop research apps.
Read the original →A new clinical study suggests Google's AMIE chatbot could support doctors by gathering patient histories and surfacing possible diagnoses before an appointment.
Read the original →SPAIA Keiba's Diamond Course members-only training-diagnosis AI will reveal three horses rated AA or higher in tomorrow's races.
Read the original →