Several big healthcare AI advances happened this week! Mediwhale's new tool uses retinal scan analysis to find heart, kidney, and eye diseases early. Instead of scary tests, it just needs a picture of your eye. This helps spot silent health issues before they become serious problems. Hospitals in Dubai, Italy, and Malaysia are already using it.

Isomorphic Labs shared that their AI-created medicines will start human trials soon. Their technology models how proteins interact to design better drugs faster than old methods. This could speed up treatments for cancer and brain diseases by finding effective compounds quicker.

Google made two powerful medical AI models free for everyone. MedGemma 27B looks at X-rays, lab slides, and health records together like doctors do. It scored 87.7% on medical tests while being cheaper than other AIs. The smaller MedGemma 4B wrote X-ray reports that real doctors said were good enough for patient care 81% of the time.

A European AI safety project began at Bambino Gesù Children's Hospital. It focuses on protecting patient information with special rules to prevent hacking. The team made patient safety their top priority when building this cybersecurity system.

Mayo Clinic developed an AI infection detector that identifies surgical wounds needing care from patient photos. It got things right 73% of the time, helping doctors catch problems faster while reducing their workload after operations.

Healthcare leaders formed a global AI office through HL7 to create fair rules for medical AI worldwide. They're already working on stopping healthcare fraud and writing guides for safe clinical AI use in hospitals.

Bioethicists warned that hospitals must get clear patient permission before using AI in treatments. They said vague forms confuse people and hurt trust. Their study pushed for transparent rules about how AIs help doctors make decisions.