Clinical & Diagnostics
Philips Verida: World's First AI-Powered Detector-Based Spectral CT Receives FDA 510(k) Clearance
Verida integrates Philips' third-generation Nano-panel Precise dual-layer detector with an AI-based deep learning reconstruction engine that reduces noise and enhances image quality across CT applications. The system reconstructs 145 images per second — enabling full exams to appear in under 30 seconds, 2× faster than prior-generation systems — and supports up to 270 exams per day. By simultaneously capturing high- and low-energy data in a single acquisition, Verida delivers conventional and spectral results together, enabling material differentiation and tissue characterization without pre-selection or repeated scanning. Cleared indications include head, whole body, cardiac, vascular, and low-dose lung cancer screening CT.
SPRY Launches First AI Agentic Scribe Built Natively for Rehab Therapy — 18x Adoption Growth, 80% Therapist Retention
Unlike passive scribes that generate a draft and stop, SPRY's Agentic Scribe remains active during the clinical session — catching documentation errors in real time, applying clinician-specific style preferences, and building on each patient's historical record to produce notes that sound like the treating provider. In five months post-launch, adoption grew 18× with no mandate. Clinics report documentation time dropping up to 75%, with revenue increasing 20–30% within 90 days. The scribe is one component of a broader autonomous platform, alongside SPRY's Prior Authorization Agent and Scheduling Agent — forming an integrated AI operations layer for outpatient rehab.
ARPA-H Launches ADVOCATE: First-Ever Program to Build an FDA-Authorized Agentic AI for Clinical Heart Care
ADVOCATE is a 39-month, two-phase program with a 3-year FDA approval target — the most aggressive federal timeline ever set for an autonomous clinical AI. Phase 1 develops a patient-facing agent capable of writing and modifying prescriptions, adjusting diet and physical therapy recommendations, and supporting diagnoses; Phase 2 includes a large-scale scalability study evaluating clinical outcomes, safety, cost-efficiency, and reimbursement pathways. A companion supervisory "overseer" AI monitors deployed agents post-launch for continued safety and efficacy. Full proposals were due April 1, 2026; ARPA-H expects to announce award teams by June 2026.
Research & Science
AI Drug Discovery Invades Longevity Medicine — And Clinicians May Not Be Ready for What's Coming
Generative AI platforms combining protein structure prediction, molecular docking simulation, and multi-omics integration are compressing the target-to-clinical-candidate timeline from years to months. More than 173 AI-discovered drug programs are now in active clinical development globally — approximately 94 in Phase I, 56 in Phase II, and 15 in Phase III — with 15–20 programs expected to enter pivotal trials in 2026 alone. The longevity medicine application layer represents a new frontier, with AI platforms now being applied specifically to aging pathway targets (mTOR, senolytics, NAD+ metabolism) as commercialization pressure accelerates.
npj Digital Medicine: AI Now Matches Expert Radiologist Performance Across Prostate Cancer Diagnostic Imaging
The review covers AI applications across transrectal ultrasound (TRUS), multi-parametric MRI (mpMRI), and PSMA PET/CT for lesion detection, Gleason grade classification, and androgen deprivation therapy benefit prediction. CNN- and transformer-based architectures achieve expert-level diagnostic accuracy across modalities, and FDA-cleared commercial tools are now available for Gleason scoring and treatment planning. Remaining barriers include dataset heterogeneity across imaging systems, limited multi-site validation studies, insufficient diversity in training cohorts, and unclear regulatory pathways for continuous-learning clinical systems.
BJC Reports: Precision Oncology in the Age of AI — From Drug Design to Clinical Translation
The paper reviews AI-driven target identification via transcriptomic and genomic multi-omics integration, generative molecular design using diffusion models and graph neural networks, and patient stratification models that combine clinical data with spatial transcriptomics. Key clinical translation milestones include AI-designed small molecules now entering Phase I/II for solid tumors, and real-world evidence models actively informing treatment selection in NSCLC, breast cancer, and colorectal cancer. The authors identify hallucination-driven target misidentification and training data quality as the primary barriers to wider clinical adoption.
Policy & Regulation
Sutter Health and MemorialCare Face Federal Class Action Over AI Scribe Recording Without Patient Consent
The complaint alleges violations of the California Invasion of Privacy Act (CIPA), Confidentiality of Medical Information Act (CMIA), Federal Wiretap Act, and California Unfair Competition Law. Plaintiffs seek class certification for a nationwide class covering patients seen at Sutter Health or MemorialCare in the past two years who had audio recorded and processed without documented consent. This follows an identical case filed against Sharp HealthCare in November 2025 involving the same Abridge platform. The lawsuits are coalescing into a wave that mirrors the early HIPAA and data-breach class action waves from the 2010s — and every health system deploying ambient scribes should review their consent workflows immediately.
Florida: Physicians Using AI Scribes Without Patient Consent Could Face Felony Charges Under State Wiretapping Law
Florida is a two-party consent state under the Florida Security of Communications Act (FSCA), which criminalizes the interception of oral communications without the consent of all parties. Unlike California's CIPA — which carries civil remedies — Florida's statute includes potential criminal penalties and felony exposure. Legal experts warn that AI scribe software recording patient-clinician conversations without documented patient consent violates this statute. Health systems have not consistently obtained such consent, and many of the verbal disclosures being used ("this visit may be recorded") may not meet the legal standard. At least 12 other states have similar two-party consent laws where parallel risk applies.
Indiana and Utah Pass Laws Directly Regulating AI Use in Health Insurance — A Growing State-Level Wave
Indiana's law (signed March 4, 2026; effective July 1, 2026) amends the state insurance code to prohibit AI-only claim downcoding without human review of the individual patient's medical record — directly targeting the automated claim reduction workflows deployed by major payers. Utah's law (signed March 19, 2026; effective January 1, 2027) requires health plan transparency reports to include whether AI tools are used in prior authorization review, enabling consumer and regulatory scrutiny of algorithmic denial patterns. These laws join at least 15 states with similar legislation pending or recently enacted, and come directly in response to CMS reporting on AI-driven claim denial rates at Medicare Advantage plans.
HHS's Clinical AI RFI Attracted 7,300+ Comments — Here's What the Industry Is Demanding
Key themes in the comment corpus include: new CPT/billing codes for AI-assisted clinical decision support to make reimbursement viable; FDA safe harbors for post-clearance updates to continuously learning AI systems to avoid re-review requirements; TEFCA data access standards extended to support AI training datasets using de-identified patient records; and liability safe harbor frameworks for health systems deploying FDA-cleared AI tools that contribute to adverse outcomes. The original RFI was issued December 23, 2025, with a February 23, 2026 deadline. As of March, only a fraction of comments had been posted publicly to the docket.
Industry & Business
STAT AI Prognosis: A $15 AI Diagnostic Test, Anthropic's Project Glasswing Expands, and Doctronic Pilot Results Surprise
The $15 AI diagnostic test leverages a lightweight model trained on diagnostic imaging and lab value combinations, designed for resource-constrained primary care settings where specialist referral is expensive or impossible. Anthropic's Project Glasswing has reportedly expanded to at least four new academic medical centers, with initial outcomes data expected mid-2026. Doctronic's pilot — in which AI autonomously handles the first 80% of a clinical encounter — showed the AI managing chronic disease follow-up at primary care visit complexity levels in approximately 23% of cases where only acute simple visits were originally expected, suggesting the clinical scope of AI-managed encounters is broader than early assumptions.
Yuzu Health Raises $35M Series A — With Anthropic's Anthology Fund — to Rebuild Health Insurance Plan Infrastructure
Yuzu's platform serves as the system of record for health plan administration — handling claims adjudication, stop-loss submissions, bookkeeping reconciliation, and downstream reporting — using AI to automate workflows that traditionally require large, expensive manual operations teams. The round was led by General Catalyst and Chemistry, with notable participation from Anthropic's Anthology Fund and Bain Future Back Ventures, and brings total capital raised to $40M. Yuzu was founded in 2022; this marks its transition from employer health plan operator to infrastructure provider serving other plans, TPA companies, and insurers. Deployment capital targets engineering expansion and national scaling.
Eli Lilly and Insilico Medicine Ink a $2.75B Global AI Drug Discovery Collaboration — the Largest of Its Kind
Insilico receives $115M upfront, with the remainder subject to clinical, regulatory, and commercial milestones plus royalties on future sales. Lilly gains exclusive worldwide rights to develop, manufacture, and commercialize novel oral therapeutics in preclinical development across multiple disease areas. Insilico's Pharma.AI platform integrates generative chemistry, protein target identification, and clinical trial outcome prediction; the company has 28 drugs in its pipeline, with nearly half already at clinical stage. This collaboration extends a prior $100M partnership signed in November 2025 and positions Insilico as the clearest example of an AI drug discovery company transitioning from platform provider to full-pipeline partner with a top-10 pharma company.
Bessemer Venture Partners: State of Health AI 2026 — AI Has Become Table Stakes, Now Comes the Execution War
Key findings: 62% of digital health VC in H1 2025 went to AI-enabled companies, with AI startups commanding an 83% funding premium per round over non-AI companies. The three highest-funded categories were non-clinical workflow automation, clinical workflow tools, and data infrastructure. Clinical decision support remains the most underfunded category relative to its potential impact. Bessemer flags three primary headwinds: AI-generated hallucinations contributing to incorrect clinical notes or billing codes, health system workflow integration failures reducing clinical adoption, and AI-enabled upcoding practices (estimated at $2B+ in excess claims per BCBS analysis) provoking payer backlash that could slow enterprise contracting.
Social Buzz
Gallup Poll: 1 in 4 Americans Now Use AI for Health Advice — and 14 Million Have Skipped a Doctor Visit Because of It
The nationally representative survey (n=5,500+ U.S. adults, October–December 2025) found that 61% of recent AI health users relied on general conversational AI tools (ChatGPT, Microsoft Copilot) and 55% used AI-augmented web search (Google AI Overviews). Top use cases: nutrition/exercise (59%), physical symptoms (58%), medication side effects (46%), and interpreting medical information (44%). Among users, trust is evenly split: 33% trust the information, 33% are neutral, and 34% actively distrust it — while continuing to use it. The rate of care-avoidance is highest among adults 18-34 and in lower-income groups, amplifying equity concerns about AI becoming a healthcare access substitute rather than a supplement.
Healthcare AI's Reckoning Has Arrived: "Prove It or Move Aside" — And the Pressure Is Real
The RadAI analysis identifies four structural shifts defining this reckoning: (1) health systems are now demanding contractual outcome commitments tied to AI tools before renewal; (2) UnitedHealth and HCA are publicly projecting AI-driven cost savings of nearly $1B and $400M respectively for 2026 — creating benchmark expectations the industry will be held to; (3) a BCBS analysis pinpointing $2B+ in AI-enabled billing inflation is triggering payer pushback that threatens enterprise contracting; and (4) hospital boards are pressing CIOs for AI ROI dashboards before approving expanded investments. The companies surviving this cycle are those with prospective outcome data, workflow-integrated deployments, and transparent performance measurement — not those still running pilots.