John Lee, MD, Emergency Physician, Edward Hospital Naperville; Informaticist, & Epic Consultant, argues health systems unlock Epic’s value by limiting customization, consolidating instances after mergers, and tightening data governance. He says many organizations use under a quarter of Epic’s capabilities and should favor Foundation build patterns, employ physician builders, and map synonyms to protect analytics. With AI and Cosmos moving insights into workflows, standardization and clean data become prerequisites, enabling decision support, benchmarking, and gains.

Eric Poon, MD, MPH, Chief Health Information Officer, Duke University Health System, outlines how Duke is formalizing AI as an operating discipline. He describes a single front door for technology requests, lifecycle oversight for AI, and pilots with hypotheses, risk tiers, and value checks. He details focus areas in inpatient operations and revenue cycle, expectations for workflow fit and maturity, and plans for AI literacy, guardrails, and monitoring of vendor-delivered features to ensure durable outcomes.

In a healthsystemCIO webinar, leaders detailed how coordinated IT training improves EHR usability and reduces clinician burnout. Gretchen Britt, Liberty Market VP Information and Technology and CIO, The University of Kansas Health System, emphasized intentional, multimodal support and specialty-specific onboarding. Clara Lin, MD, VP/CMIO, Seattle Children’s, highlighted at-the-elbow coaching, in-workflow micro-lessons, and knowledge platforms. Dirk Stanley, MD, CMIO, UConn Health, framed training as deliverable with owners, metrics, and toolkits to drive adoption and ROI.

Michael Westover, VP, Data Partnerships & Informatics, Providence, argues healthcare data reciprocity remains lopsided: providers supply rich clinical data while payers return claims stripped of financial and sensitive behavioral health elements, leaving providers with limited operational insight. He urges providers to advocate for parity as the 2027 Provider Access API advances but falls short. Westover outlines payer data domains—benefits, out-of-pocket status, network details, policies, census, longitudinal records, SDoH, formulary, and value-based contract metrics and terms.

Health systems reshape ambulatory care by centralizing governance, via hub-and-spoke operations, and shifting volume from inpatient care to clinics, virtual services, and home. Leaders target growth in clinics, ASCs, cardiology and GI while pruning underperformers. Partnerships with physician groups and technology firms, including AI tools, extend capacity. Enterprise scheduling improves throughput. Performance is tracked via financial, clinical, access, experience, and retention metrics, according to the Center for Connected Medicine (CCM) at UPMC and KLAS.

Michael Weiss, MD, Epic Physician Builder – Associate CMIO at Memorial Healthcare System, describes the “Rogue Note Initiative,” created to balance clinician EHR personalization with standardized, structured documentation for quality, safety, and revenue. After identifying critical elements and tethering them to core note sections—backed by interruptive decision support—the initiative discouraged unmaintained note variants. Within weeks, rogue notes declined from 487 of 6,078 weekly notes to 24, a ~95% reduction over 20 weeks, while preserving flexibility.

Developed by the Health AI Partnership, a multi-stakeholder collaborative, the HAIP AI Vendor Disclosure Framework helps healthcare delivery organizations evaluate AI systems responsibly. It guides vendors and HDO teams through five domains: system capabilities and intended use; performance and compliance; data stewardship; integration requirements; and lifecycle management. The framework supports procurement and implementation by standardizing disclosures, clarifying accountability, assessing risks and biases, and illuminating total cost of ownership for safe, effective, ethical adoption and governance.

Sarah Gebauer, MD, Senior Physician Researcher, RAND, proposes the FAILURE Framework for clinical AI red-teaming, drawing on patient-safety lessons that system errors follow patterns. FAILURE spans Framing & Feedback, Attention & Automation, Interaction & Interface, Learning & Latency, Uncertainty & Under-specification, Responsibility & Role Clarity, and Environment & Emergence. The piece highlights human-in-the-loop pitfalls, interaction harms, and sociotechnical dynamics, urging exercises that test governance and workflows before deployment, so AI systems fail safely, not silently.

HSCC's Cybersecurity Working Group released the Health Industry Cybersecurity Sector Mapping and Risk Toolkit (SMART) to help healthcare organizations map third-party dependencies and quantify systemic risk across clinical, administrative, and manufacturing workflows. Built over 16 months by 80 organizations, the resource aligns with National Cybersecurity Awareness Month and HSCC’s 2024–2029 plan. Leaders warn ransomware and outages create one-to-many impacts. SMART helps smaller entities strengthen supplier security and craft contingency plans.

KLAS’ Credentialing 2025 report finds vendor relationships and proactive service correlate with satisfaction and efficiency. ASM and Modio Health are praised for hands-on engagement, while RLDatix and QGenda provide robust status-tracking dashboards. Full automation remains elusive; organizations rely on spreadsheets. Usability varies, with symplr and HealthStream facing clinician-workflow challenges. ASM and Modio Health drive tangible outcomes and reduced manual work.