Closing Care Gaps Autonomously in Healthcare
A regional health system deployed a Clinical Workflow Agent to autonomously identify and close preventive care gaps—improving HEDIS scores and reducing the burden on care coordinators by 60%.
The Challenge
“Care gap closures—ensuring that patients receive recommended preventive screenings, vaccinations, and chronic disease monitoring—are a critical quality metric for health systems but a labor-intensive process. Care coordinators manually reviewed patient registries, identified gaps, drafted outreach communications, scheduled appointments, and tracked follow-through. With a panel of 100,000+ members, the coordination team was overwhelmed: only 42% of identified care gaps were being closed within measurement periods, impacting HEDIS quality scores and value-based contract performance.”
The Solution
Eficens deployed a Clinical Workflow Agent integrated with the health system's Epic EHR and secure messaging platform. The agent continuously monitors the patient registry for care gaps, prioritizes outreach based on clinical urgency and predictive engagement scores, drafts personalized outreach messages, and routes them to the appropriate channel (secure portal message, automated call, SMS) based on patient preferences and prior engagement history.
Implementation
EHR Integration and Registry Monitoring
The Clinical Workflow Agent integrates with Epic via the FHIR R4 API, subscribing to patient data updates and querying the care gap registry daily. A HIPAA-compliant data processing layer handles all PHI in a HITRUST-certified environment, with Governance Gate redaction ensuring that PHI never leaves the secure clinical data boundary. The agent's monitoring rules are configured against the health system's specific HEDIS measure definitions, ensuring alignment between agent-identified gaps and the measures that affect quality scores.
Personalized Outreach Generation
For each identified care gap, the agent generates a personalized outreach message incorporating the patient's name, their specific care gap (e.g., "It has been 14 months since your last HbA1c test—our guidelines recommend testing every 12 months for patients with Type 2 Diabetes"), available appointment slots, and the care coordinator's contact information. Message generation follows approved clinical language templates reviewed by the health system's clinical informatics team, preventing the agent from generating novel clinical recommendations.
Scheduling and Follow-Through Tracking
When patients respond to outreach, the agent handles appointment scheduling by accessing Epic's scheduling API—presenting available slots and creating the appointment upon patient selection. A follow-through tracking module monitors whether scheduled appointments are kept and, for patients who do not respond to initial outreach, automatically sends two follow-up messages at 14 and 30 days. Patients who remain unresponsive after three contacts are flagged for direct care coordinator outreach, with the full contact history attached to the patient record.
Related Resources
View allBuilding Reliable RAG Architectures for Regulated Industries
Healthcare, financial services, and legal sectors demand more than accurate AI—they require verifiable, auditable, data-resident AI. RAG architectures purpose-built for regulated industries look very different from general-purpose implementations.
The "Governance Gate": How We Redact PII and PHI by Default
In enterprise AI, data protection cannot be an afterthought. The Governance Gate is an architectural pattern that intercepts every agent action, identifies sensitive data, and redacts it before it reaches external systems—by default.
Enterprise AI Governance: A Framework for PHI/PII Protection
A regulatory-grade framework for protecting sensitive data across the full AI pipeline—from ingestion through retrieval, reasoning, and action—with implementation guidance for HIPAA, GDPR, and CCPA compliance.