AI Laws, Clinical Results & Tech Deals Shape Healthcare Oct 7

Today marks another day of AI reshaping healthcare operations, regulations, and investment. From Pennsylvania's new legislation to impressive diagnostic accuracy results, the pace of change continues to accelerate. Here's what busy healthcare professionals need to know:

Have suggestions? Reply to this email.

Pennsylvania lawmakers introduced bipartisan legislation to regulate AI use in healthcare. The bill covers clinical and administrative settings as part of a broader state push for AI oversight. Source Hospitals and vendors will face new compliance requirements around procurement, documentation, and risk management. Legal teams should track this bill's progress and prepare policy updates. Procurement processes should add AI governance checks for vendor selection. Clinical leaders need to plan for new auditing and transparency requirements.

A literature review shows AI models outperforming clinicians on specific tasks. For IgA nephropathy, an artificial neural network predicted disease progression with 87% accuracy compared to 69.4% for nephrologists. Source Other results include 98.3% accuracy for urinary tract infection diagnosis and 95% accuracy for predicting survival in kidney cancer patients. These gains come with training and governance needs. Start with narrow, measurable use cases. Plan upfront training, data privacy, and traceability to meet FDA requirements. Source

Researchers developed a hierarchical reasoning model (HRM) using only 27 million parameters. It outscored much larger models on reasoning tests: 40.3% vs OpenAI's 34.5% and Claude's 21.2%. Source The model uses brain-inspired design with a slow planner and fast solver. Source Smaller, efficient models could cut compute costs for healthcare AI. The team open-sourced the code for testing. Source However, experts suggest training methods, not just architecture, drove the gains.

AI scribes do more than cut charting minutes. They change how clinicians work mentally. Physical medicine doctors spend over 19 hours weekly on admin tasks. Source AI scribes transcribe visits into notes and plans, improving accuracy and reducing mental burden. Source Practice leaders report better work-life balance and higher productivity per evaluation, not just time saved. Key needs include native EMR integration and ambient transcription for noisy clinics. Measure both time saved and clinician mental load when evaluating tools.

Qualtrics agreed to acquire Press Ganey for $6.75 billion in cash and private stock. Source Press Ganey serves over 41,000 hospital systems. The deal combines Qualtrics' AI survey tools with Press Ganey's extensive hospital data. Source Eleven banks and private firms are providing debt financing. Hospitals should expect new AI tools built on patient datasets and potentially tighter vendor relationships. This marks a major bet on health data as AI fuel.

OpenAI signed a multi-year chip deal with AMD, planning 1 gigawatt of AMD Instinct MI450 chips by late 2026, potentially expanding to 6 gigawatts. Source The deal includes warrants for up to 160 million AMD shares tied to performance targets. AMD stock jumped over 30% on the news. More chip suppliers reduce single-vendor risk and could push down AI compute prices. Lower costs accelerate AI adoption in healthcare imaging, diagnostics, and drug discovery. Nvidia still supplies up to 10 gigawatts to OpenAI, so this adds competition rather than replacing existing deals.