Clinical Trials Get an AI Redesign
Federal health agencies are testing an AI-enabled approach to clinical trials, while biopharma and healthcare organizations are bringing AI into R&D, commercial planning, nursing and patient support. Epic’s discovery of security weaknesses in its systems also shows why access controls need to keep pace. Outside healthcare, AI agents are starting to complete transactions and work across business systems.
Clinical Trials Get a New Architecture
What’s Happening
On September 30, HHS and ARPA-H launched SURPASS, a program designed to combine predictive models, shared control groups, real-time analysis and agentic operations in continuous adaptive platform trials. Three companion projects target faster site activation, privacy-by-design consent and data infrastructure, and patient navigation. The announcement describes intended tools and frameworks; it did not specify program funding.
Why This News Is Important
The program goes beyond automating individual trial tasks. It brings trial design, operations, data access and patient participation into one system, putting AI closer to the work that shapes development, regulatory decisions and launch timing.
What It Could Mean
If the methods earn regulatory confidence, sponsors could eventually test more questions within shared infrastructure, adapt sooner as evidence accumulates and reduce repetitive startup work. Commercial and Medical Affairs teams may also gain earlier, more continuous visibility into emerging evidence, although the program still has to prove rigor in practice.
Source: ARPA-H
Read more: HHS launches SURPASS and new efforts to accelerate faster, smarter clinical trials
CSL Connects AI to Regulated R&D
What’s Happening
CSL announced a collaboration with AWS to apply cloud and AI tools across research and clinical development. The companies plan to connect scientific data, laboratory work and computational models, while reducing manual effort in protocol authoring, clinical-data handling and regulatory submissions. CSL says the work will retain data governance, traceability and inspection readiness, and include AI training for R&D staff.
Why This News Is Important
CSL is applying AI across connected scientific and document workflows rather than treating it as a separate discovery experiment. For Medical Communications and regulatory teams, traceability and inspection readiness need to be built into the workflow from the start.
What It Could Mean
This approach could make it faster to carry a new scientific finding into the documents, evidence packages and internal decisions it affects. It could also make it easier to reuse structured evidence across medical content, as long as experts remain responsible for interpretation and final submissions.
Source: CSL
Read more: CSL and AWS collaborate to advance biopharmaceutical R&D
Commercial Targeting Gets More Clinical Context
What’s Happening
AcuityMD added prescription data to AcuityAI, combining it with procedure and diagnosis information in a single workflow for medtech commercial teams. Representatives can query prescribing patterns and identify HCPs treating patients whose therapies may indicate a device opportunity. The company cited current customer uses, but the announcement did not provide independent evidence of sales or patient-outcome improvement.
Why This News Is Important
Commercial teams are getting tools that let them look beyond static target lists and ask more specific questions about treatment patterns. That can make HCP engagement more relevant, but teams still need strong data governance and care when turning population-level patterns into individual outreach.
What It Could Mean
For pharma brands, similar tools could support better segmentation, field planning and next-best-action recommendations based on treatment patterns rather than channel activity alone. The value is better context for conversations, not automated assumptions about a clinician or patient.
Source: AcuityMD via Business Wire
Read more: AcuityMD brings prescription data into AcuityAI
Nurses Get Their Own Evidence Assistant
What’s Happening
Elsevier launched ClinicalKey Nursing AI globally, with customer access rolling out during 2026. The natural-language tool is designed to answer nursing questions about procedures, medications and discharge planning using Elsevier content and material from nursing organizations, journals, guidelines and clinical-skills resources. Answers include citations, and the product is positioned to support rather than replace nurses’ clinical judgment.
Why This News Is Important
Most clinical assistants have been designed around physicians. A nursing-specific product reflects the fact that nurses have different information needs and workflows, and it gives them cited answers they can check.
What It Could Mean
Medical Affairs and education teams may need to think about how scientific information reaches nurses through AI tools, not just physicians. Better-structured nursing evidence could support discharge education and patient conversations, but adoption will depend on local governance, up-to-date content and trust at the point of care.
Source: Elsevier
Read more: Elsevier launches ClinicalKey Nursing AI
The Patient Relationship Extends Between Visits
What’s Happening
Welldoc introduced an AI-powered “care loop” for people managing complex conditions such as diabetes, obesity and early-stage dementia. Its Simon agent is intended to turn daily health data into personalized guidance and surface insights for care teams between visits. Welldoc attributes the system to closed, healthcare-focused small language models; engagement and deployment figures in the announcement are company-reported.
Why This News Is Important
The model treats AI as an ongoing part of care between visits rather than a stand-alone chatbot. That matters for adherence and support because personalized guidance needs current information and a clear handoff to a clinician when needed.
What It Could Mean
Pharma patient-support programs could eventually become more responsive to day-to-day needs, with education, reminders and human outreach changing as circumstances change. Any such model would require clear consent, careful boundaries around medical advice and evidence that more engagement produces better outcomes.
Source: Welldoc
Read more: Welldoc launches its AI-powered care loop
Virtual Care Moves Inside the Health System
What’s Happening
UVA Health and K Health announced a strategic partnership to expand virtual primary care and behavioral-health access across Virginia. K Health’s clinical AI infrastructure is expected to support the visit experience inside UVA Health’s patient portal, with administrative assistance intended to leave more time for direct care. The partners did not report rollout scale or clinical outcomes.
Why This News Is Important
AI-enabled virtual care is moving inside established health systems. Patients may increasingly expect virtual care to feel like part of the health system they already use, rather than a separate app or service.
What It Could Mean
Pharma teams could find more patient education, support and referral moments occurring inside health-system-owned virtual journeys. That could increase the value of services and content that fit easily into provider workflows, preserve trust and hand difficult questions to licensed clinicians.
Source: UVA Health and K Health via Business Wire
Read more: UVA Health and K Health announce virtual-care partnership
AI Finds the Risk Behind the Record
What’s Happening
The New York Times reported that Epic used an Anthropic AI tool to uncover security weaknesses that could have allowed undetected access to patient data. According to the report, the severity and unusual nature of the vulnerabilities led Epic to slow development of some product lines while engineers worked on patches. The report did not say the flaws had been exploited.
Why This News Is Important
AI can help find security weaknesses, but the Epic case also shows how much risk sits inside connected health-data systems. As AI agents gain access to records and workflows, access controls, activity logs and monitoring become patient-trust issues, not just technical safeguards.
What It Could Mean
Pharma and healthcare partners may face more scrutiny before connecting AI agents to patient, HCP or clinical systems. Security reviews may increasingly test what an agent can infer, access and do across connected tools, with clear, reviewable permissions becoming as important as answer accuracy.
Source: The New York Times
Read more: Medical records firm uses AI to expose flaws that threaten patient privacy
Outside the Industry
Shopping Agents Reach the Checkout
What’s Happening
Shopify added WebMCP support for checkout, allowing browser-based AI agents to inspect and update a checkout and submit an order after the buyer authorizes it. The tools work with Shop Pay and are rolling out to eligible merchants. Shopify says the structured interface is designed to preserve accurate product facts, required disclosures and handoff requirements.
Why This News Is Important
AI shopping agents are moving from helping customers find products to completing purchases. That could make it normal for an assistant to finish a task, as long as the customer approves the final step and required disclosures are clear.
What It Could Mean
Healthcare is more constrained, but people may come to expect similar experiences. Patients could come to expect assistants that move from information to verified next steps such as enrollment, appointment booking or support-program navigation, while asking for consent and handing off to the right person when required.
Source: TechCrunch
Read more: Shopify opens checkout to browser-based AI agents
Business Agents Learn the Whole Operation
What’s Happening
Meta expanded Muse to small businesses with connectors for Facebook and Instagram business accounts plus tools including Shopify, Slack, Stripe, Canva, QuickBooks and Klaviyo. Meta says the agent can use brand, storefront, financial and customer context to complete work, while requiring approval before anything is published, sent or spent. Usage is free within limits.
Why This News Is Important
Business AI is starting to work across multiple systems instead of one task at a time. That gives the agent more context about products, customers, brand voice and campaign performance.
What It Could Mean
Pharma may eventually use similar assistants across approved content, customer data, planning and measurement. The challenge will be deciding what information each agent can use and where medical, legal, regulatory or other human approval is required.
Source: Meta
Read more: Introducing Muse for Small Business
Summary
This week’s stories show AI moving into more of the work around clinical trials, R&D, commercial planning, nursing and patient support. Epic’s security findings are also a reminder that access controls and monitoring need to keep pace. Outside healthcare, Shopify and Meta are showing how AI agents may move from answering questions to completing tasks across connected systems.