Healthcare AI News 3/19/25

News

Nvidia releases GR00T N1, an open source foundation model for generalist humanoid robots.

Google unveils new health-focused features at its The Check Up healthcare event:

• What People Suggest, a new search feature that summarizes online discussions from individuals with the same condition.
• FHIR support for Health Connect, which is now API-enabled to read and write medical data in FHIR format.
• Loss of pulse detection, an FDA-cleared Pixel Watch 3 feature that automatically calls emergency services if the wearer’s heart stops.

Korea-based LG AI Research announces Exaone Deep, an LLM with advanced reasoning capability that the company says beats DeepSeek R1 on math benchmarks while being 95% smaller.

Researchers find that Metat’s open source AI model Llama outperformed OpenAI’s GPT-4 in diagnosing complex medical cases. Open source LLMs offer healthcare advantages such as auditability, bias detection, and domain-specific tuning for medical literature and patient records. They integrate more easily into healthcare systems, reduce costs, prevent vendor lock-in, and can be deployed locally or in secure environments to protect patient data.

Illinois lawmakers advance bills to restrict online behavioral health providers from using AI for therapy sessions and to limit insurer use of AI to deny coverage.

India’s Apollo Hospitals will increase its investment in AI with the goal of freeing up 2-3 hours daily for doctors and nurses. AI is being used to suggest diagnoses, recommend tests and treatments, transcribe physician notes, and generate daily schedules from nurse documentation.

Business

R1 and Palantir launch R37, an AI lab that will focus on developing automation solutions for healthcare reimbursement.

Urgent care operator CityMD will use Notable’s AI platform and agents for patient scheduling, messaging, registration and intake, and payments. CityMD is owned by VillageMD, whose majority owner Walgreens Boots Alliance is expected to seek a buyer for the business after its own acquisition by a private equity firm.

Other

A JAMA Viewpoint article argues that since lawmakers are unlikely to expand the FDA’s oversight of AI standards, private governance mechanisms — such as licensing agreements between AI developers and health systems — will be necessary. The authors also warn that litigation over AI-caused patient harm will rise, but that won’t drive safety improvements because physician users will bear the brunt of liability.

Contacts

Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Contact us.