|
AI is reshaping healthcare—fast. This brief cuts through the noise for health leaders: 3 signals from the market, 2 research studies, 1 insight to act on. Curated weekly.
|
|
This week: A class action puts AI claim denials on trial, every major payer announces AI initiatives, and states push back against federal preemption. Plus, new research on AI "digital twins" for surgical decisions—and whether half of U.S. hospitals are ready for GenAI.
|
A class-action complaint alleges Humana used nH Predict—an AI tool owned by UnitedHealth Group—to wrongfully deny Medicare Advantage members' medical claims. The lawsuit claims the algorithm systematically overrode clinical judgment to reject coverage. UnitedHealth acquired nH Predict through its 2020 purchase of naviHealth, and now licenses it to other insurers including Humana.
So what?
The legal theory: if your AI vendor's tool is systematically denying claims, you're both liable.
Read the full story →
Becker's year-end roundup: UnitedHealth hired a Chief AI Scientist. Cigna launched AI chatbots and plan-selection tools. Optum unveiled "Optum Real" for instant claims validation. Elevance's virtual assistant now reaches 22 million members, with Medicare expansion planned for 2026. Aetna rolled out a generative AI assistant. Oscar announced Oswell, powered by OpenAI.
So what?
AI went from pilot to production across every major payer in 2025. The differentiation now is execution, not adoption.
Read the roundup →
Manatt's year-end policy tracker: 47 states introduced over 250 AI bills affecting healthcare in 2025. Thirty-three were enacted across 21 states. California banned companion chatbots without suicide-prevention protocols. Colorado expanded AI governance requirements to health insurers. Pennsylvania proposed mandatory AI disclosure in clinical communications. Then on December 11, President Trump issued an executive order directing agencies to challenge state AI laws deemed "overly burdensome"—including potential withholding of federal funds.
So what?
States are regulating faster than Washington. The EO signals friction ahead, but only Congress can preempt state law.
Read the policy tracker →
|
A randomized controlled trial tested Joint Insights, an AI tool co-developed by OM1 and Dell Medical School that generates patient-specific simulations ("digital twins") predicting surgical benefit, risk, recovery, and quality-of-life for knee osteoarthritis patients. Results: Patients using the AI showed higher decision quality, better medium- and long-term physical function, and lower decision conflict compared to digital education alone.
Why it matters
Surgical appropriateness is a $10B+ problem. And nearly 20% of knee replacements don't meaningfully improve outcomes. AI that predicts who won't benefit—before the procedure—changes the economics.
Read the study →
Generative AI—which creates novel text, images, or audio—has emerged as a practical tool in healthcare: ambient scribes, billing code suggestions, and draft patient messages are now in production. A survey study found that by end of 2025, 50% of nonfederal acute care hospitals will use these tools. However, the study flags a "digital divide"—some hospitals are adopting without complete safeguards or governance frameworks in place.
Why it matters
Adoption is outpacing infrastructure. The hospitals moving fastest aren't necessarily moving safest.
Read the study →
|
Menlo Ventures' 2025 State of AI in Healthcare report identifies ambient scribes as the industry's first true breakout application: $600 million in revenue this year, up 2.4x year-over-year.
The category minted two new unicorns—Abridge (30% market share) and Ambience Healthcare. In total, AI scribe companies have raised over $1.5 billion in the past 18 months alone, including Abridge's $300M round, Ambience's $243M raise, and Microsoft's DAX Copilot rollout across 650 health systems.
Why scribes? The ROI is immediate and measurable: documentation time drops, physician satisfaction rises, and—critically—the data captured creates a moat for future AI applications.
Takeaway
Scribes aren't the endgame. They're the data layer for what comes next—clinical decision support, coding, quality measurement. The companies winning scribes are very likely positioning for the full clinical AI stack.
Read the Menlo report → | Forbes analysis →
|
|