A 2026 meta-analysis of 42 studies and nearly 35,000 patients confirms AI-assisted colonoscopy finds more adenomas and misses fewer polyps, but wide prediction intervals and signs of publication bias mean the benefit isn't consistent everywhere — and exams take slightly longer.
A July 2026 field report from Ethiopia's Eliminate TB Project shows AI chest X-ray screening, now live in 225 health facilities, found 84% of all nationally detected TB cases despite covering just 68% of people screened — with hit rates varying sharply by population group.
At AACR 2026, Natera, MD Anderson, and a multi-institution NCI-led team showed AI models that predict cancer mutations and immunotherapy response directly from routine pathology slides, with Natera reporting 98% accuracy on a key colorectal biomarker — though none of it is peer-reviewed yet.
UpDoc has won FDA clearance for a patient-facing large language model that gives adults with type 2 diabetes voice- and text-based insulin titration guidance with no mandatory physician review step, a first that is already drawing as much scrutiny over governance as praise for the technology.
Bayesian Health's TREWS system, built on Johns Hopkins research, has received FDA 510(k) clearance as what the company says is the first AI tool cleared specifically to flag sepsis before a clinician suspects it, citing a study of over 764,000 patient encounters.
The 2026 NCCN breast cancer screening guidelines now incorporate AI-derived risk scores from standard mammograms, using a 5-year risk threshold of 1.7 percent to trigger shared decisions about additional screening like MRI.
An IAPP privacy analysis of five health-focused AI agents launched within three months found each promises not to train on users' health data and keeps health chats isolated from ordinary conversations, but verification of those promises remains out of reach for outside researchers.
OpenAI's January 7, 2026 launch of ChatGPT Health lets users link medical records and wellness apps to the chatbot, with what the company describes as purpose-built encryption and isolation designed specifically for health conversations.
A Kentucky federal lawsuit, Barrows et al. v. Humana, accusing the insurer of using AI tools to improperly terminate post-acute care coverage for Medicare Advantage patients, received a scheduling order in July 2026 and remains actively litigated.
West Virginia University Health System has expanded its Abridge AI documentation platform to more than 2,800 clinicians across 25 hospitals, with internal surveys showing a 61 percent reduction in clinicians' self-reported cognitive load.