Practical AI Now: Governance, Risk and ROI at the HIMSS AI Leadership Strategy Summit
HIMSS leaders urge practical AI: start with narrow, high-value use cases, integrate into workflows, and measure ROI. Build governance for HIPAA, safety, bias, and ongoing review.

HIMSS AI Leadership Strategy Summit: Deploy AI, Manage Risk, Drive ROI
Chicago sent a clear signal: be prepared for change when implementing AI. The focus of the two-day summit is practical-deploy AI that improves operations and outcomes while keeping risk, HIPAA, and patient safety in check.
"We're at a critical moment in time to redesign ourselves," said HIMSS President and CEO Hal Wolf. Many leaders are ready to apply AI to administrative inefficiencies, clinical decision support, and measurable health outcomes-if they can lock in governance, ROI, and change management.
Wolf put it plainly: "The technology for learning, that's the easy part. How we bring it into the stream is the hard part." His caution to executives: NT + OO = COO-New Technology plus Old Organization equals Costly Old Organization.
What executives heard
- AI can address staffing shortages and an aging population by predicting demand and identifying high-risk patients for proactive management.
- Healthcare isn't short on data; it's short on usable information and institutional knowledge. Data must be transformed, governed, and operationalized.
- Governance isn't a committee-it's a system. Think accountability, model oversight, HIPAA alignment, and patient safety baked into every stage.
- ROI is earned through process change, not pilots alone. Failed initiatives usually lacked adoption, measurement, or fit with core workflows.
From intent to implementation: a practical path
- Define narrow, high-value use cases: Prior authorization automation, no-show prediction, readmission reduction, discharge optimization.
- Set governance up front: Risk classification, approval gates, model documentation, bias testing, PHI controls, and human-in-the-loop requirements.
- Integrate into the "stream": Build into existing workflows, not around them. EHR integration, routing, and alert thresholds matter more than model accuracy on paper.
- Measure and manage ROI: Baseline your KPIs, run A/B pilots, quantify lift, and tie gains to cost, quality, throughput, and experience.
- Plan for adoption: Training, change champions, SOP updates, and clear escalation paths. No adoption, no ROI.
- Scale with discipline: Post-implementation monitoring, drift detection, periodic revalidation, and version control for models and prompts.
Data to decisions
Healthcare data is doubling every 12-18 months. Volume isn't the advantage-conversion is. As Wolf noted, data must become information, then knowledge, or it sits idle in systems.
- Data readiness: Inventory sources, assess quality, and reconcile definitions for core metrics.
- Privacy and security: Minimize PHI exposure, log access, and enforce least-privilege. Keep a clear line between experimentation and production.
- Interoperability: Standardize interfaces and event routing so insights land where decisions are made.
- Feedback loops: Capture outcome labels to improve models and update prompts with real-world performance.
Mindset and urgency
Isaiah Nathaniel, senior vice president and CIO of Delaware Valley Community Health, urged leaders to keep a growth mindset. The key questions: "What would I like to get out of it? What am I trying to do with AI? I want to get it into practical use now."
That mindset shifts AI from experiments to outcomes. Choose one operational win, one clinical support use case, and one safety gate-and ship them with governance, measurement, and adoption built in.
Governance and compliance resources
- NIST AI Risk Management Framework for risk classification and controls.
- HHS HIPAA Security Rule for safeguards and policy alignment.
Next step
Stand up a cross-functional AI steering group, lock in two production pilots with clear KPIs, and publish a 90-day rollout plan that ties technology to workflow change. Keep the formula in view: avoid "Costly Old Organization" by pairing new tech with new ways of working.
If you're building executive capability and team skills for these use cases, see curated AI courses by job role for structured, practical upskilling.