|
3, 2, 1: Health AI Brief
Every Friday
March 27, 2026
|
|
|
AI is reshaping healthcare fast. Below are 3 key AI developments, 2 studies, and 1 takeaway for this week to help you better lead with AI. Target read time: 5 minutes. |
|
UnitedHealthcare launched Avery this week — a GenAI companion that helps members navigate coverage, schedule appointments, estimate costs, and manage benefits. Currently live for 6.5 million employer-sponsored and 160,000 Medicare Advantage members, UHC plans to expand access to 20.5 million commercial, Medicare, and Medicaid members by year-end. When Avery can't resolve an issue, it transfers the member to a human advocate with a summary of the conversation. The tool is part of UHC's $1.6 billion AI investment this year, joining over 1,000 AI-powered tools the insurer currently offers. So what?
The largest insurer in the country is putting a GenAI companion in front of 20.5 million members this year. Benefits navigation via AI is quickly moving from differentiator to table stakes. Doctronic closed a $40 million Series B (co-led by Abstract and Lightspeed Venture Partners), bringing total capital to $65 million across 3 rounds in under a year. The New York-based startup offers a free AI doctor consultation platform with over 1 million users and 15 million medical conversations to date. The milestone: in December 2025, Doctronic became the first AI to autonomously renew prescriptions under Utah's regulatory sandbox, covering 190 medications with strict safety and reporting protocols. Meanwhile, Rep. Schweikert's "Healthy Technology Act" (HR 238) would formally recognize AI as a licensed practitioner eligible to prescribe FDA-approved drugs — it's pending review by the House Committee on Energy and Commerce. So what?
AI is legally renewing prescriptions in Utah, and a federal bill would recognize AI as a licensed practitioner nationwide. The prescribing threshold has been crossed; the question now is how fast everyone else follows (or not). Qualified Health, a public benefit corporation providing a secure enterprise AI platform for health systems, raised $125 million in Series B funding led by New Enterprise Associates. The platform supports over 500,000 users across Mercy, Emory Healthcare, Jefferson Health, and 8 University of Texas System institutions — including MD Anderson and UT Southwestern. Partners represent approximately 7% of U.S. hospital revenue. UT Medical Branch generated $15 million in measurable run-rate impact within 6 months of deployment. The platform provides workflow automation, agent development, clinical safeguards, and governance infrastructure. So what?
$15 million in measurable impact in 6 months with 1 health system is impressive ROI. But what Qualified Health is selling isn't just AI — it's the governance infrastructure around it. The latter is a bigger deal. |
|
The LungIMPACT trial — a prospective, multicenter RCT across 5 NHS trusts — tested whether AI prioritization of primary care chest X-rays reduced time to CT scan and lung cancer diagnosis. The study analyzed 93,326 X-rays, randomized by day: 45,987 with AI prioritization on, 47,339 with it off. Median time to CT scan was 53 days in both groups (P = 0.31). Among 558 lung cancer diagnoses (0.6% of X-rays), median time to diagnosis was 44 days vs. 46 days (P = 0.84). No significant differences in urgent referral time, treatment initiation, or disease stage. The authors concluded: "CXR AI deployments should not include worklist prioritization in this context." Why it matters
The AI worked — it flagged findings correctly. But the pathway didn't move. The 53-day wait wasn't caused by reading speed; it was caused by CT scanner availability, scheduling, and referral capacity downstream. A faster flag doesn't compress a wait time that's driven by something else entirely. KFF's latest Tracking Poll on Health Information and Trust surveyed 1,343 U.S. adults (February 24 – March 2, '26). 32% reported using AI chatbots for health information in the past year — equal to the share who use social media for health. 41% of AI users uploaded personal medical records to chatbots. Among those who used AI for mental health questions, 58% never followed up with a doctor or other health professional; for physical health, 42% didn't follow up. Adults under 30 are 3.5x more likely than those 50+ to use AI for mental health (28% vs. 8%). Uninsured and lower-income adults disproportionately cited cost and access barriers as their reason for turning to AI. Why it matters
For most users, AI is a convenience. For uninsured and lower-income adults, it's a workaround — cost and access barriers are what's pushing them there. And more than half of mental health users never follow up with a clinician at all. |
|
The Algorithm Worked. The System Didn't.
Advocate Health called over 15,000 hypertension patients in 12 days using conversational AI from Hippocratic AI. About 1,200 were cleared with normal BP. 274 with abnormal readings were connected to care. Another 400 were flagged for unrelated health issues. The AI created what Advocate's VP of AI called "a net-new access point" — reaching patients who hadn't been rechecked in over a year. Meanwhile, the LungIMPACT trial in Nature Medicine found the opposite. AI-driven prioritization of 93,326 chest X-rays did nothing to speed lung cancer diagnosis — median time to CT was 53 days, AI on or off. The reading was never the bottleneck. Scanner availability, scheduling, and referral capacity were. Takeaway
As AI scales across healthcare, the AI that worked created new capacity. The AI that didn't accelerated a step where speed wasn't the actual constraint. This distinction matters — not everything is ready to be accelerated with AI. |
|
|
Know someone who'd find this useful? Share |
The Algorithm Worked. The System Didn't.
3 signals, 2 studies, 1 insight for health leaders
Mar 28, 2026
•
1 min read