Kintsugi, a Berkeley AI startup that raised $30 million to detect depression from voice patterns, shut down commercial operations in February 2026 after years in FDA review and open-sourced its technology on Hugging Face.
MyFitnessPal acquired Cal AI, a photo-based calorie-counting app built by 19-year-old Zach Yadegari that generates over $30 million in annual revenue and has logged more than 15 million downloads in under two years.
Google retired the Fitbit app brand on May 19, 2026, relaunching it as Google Health with a Gemini-powered AI coaching assistant bundled into a new $9.99-a-month Premium tier and a $99.99 screenless Fitbit Air tracker.
Boston Children's Hospital researchers used OpenAI's o3 Deep Research model to reanalyze 376 unsolved pediatric rare-disease cases, confirming 18 new diagnoses in a study published in NEJM AI on June 18, 2026, while outside experts caution that AI-generated candidate diagnoses still require rigorous human verification before wider clinical use.
A Minnesota federal magistrate judge ordered UnitedHealth to disclose internal records on its nH Predict claims-denial algorithm in the Estate of Gene B. Lokken class action, as Alabama's new law restricting AI-only insurance denials takes effect October 1, 2026.
Sleep Cycle is validating an AI tool that screens for obstructive sleep apnea using only an iPhone's microphone, targeting the estimated 950 million adults worldwide with the condition, nearly 80% of whom remain undiagnosed.
Oura's new $399 Ring 5 adds blood pressure signals, GLP-1 tracking, and an AI-plus-physician partnership with Counsel Health, as the $11 billion company bets on clinical credibility beyond sleep tracking.
An Imperial College London AI model reads routine ECGs in under two seconds, detecting heart failure in up to 81% of cases and aortic valve disease in up to 90%, and is now being trialled on 590 NHS patients across London and Bristol.
Northwell Health will deploy Abridge's ambient AI documentation platform across all 28 of its hospitals and roughly 20,000 physicians, citing a study showing up to 67% reductions in documentation-related burnout, even as separate research finds AI-generated clinical notes contain errors at a higher rate than physician-written ones.
The FDA has cleared TREWS, the first continuous AI sepsis monitor, after a Nature Medicine study of 764,707 patient encounters found an 18% relative drop in mortality when clinicians acted on its alerts, which can flag sepsis up to 48 hours before clinical suspicion.