Get in Touch

Course Outline

Foundations of Agentic AI for Healthcare

  • Distinctions between agentic and tool-only LLM applications.
  • Autonomy boundaries, policies, and human oversight.
  • Healthcare data landscape and constraints (EHR, FHIR, PHI).

Designing Agent Workflows

  • Planning, memory, tool use, and reflection loops.
  • Prompt engineering, functions/tools, and action selection.
  • State management and orchestration patterns.

Retrieval-Augmented Agents

  • Medical document ingestion and chunking.
  • Embeddings, vector stores, and relevance evaluation.
  • Grounding responses and citation strategies.

Healthcare Integrations and Interoperability

  • FHIR/SMART basics for agent connectivity.
  • Working with structured and unstructured clinical data.
  • Eventing, APIs, and audit trails.

Safety, Risk, and Governance

  • Guardrails, red-teaming, and fail-safe design.
  • PHI handling, de-identification, and access controls.
  • Human-in-the-loop review and escalation paths.

Evaluation and Monitoring

  • Offline evaluations, golden sets, and KPI definition.
  • Hallucination detection and factuality checks.
  • Observability, logging, and cost/latency management.

Deployment Patterns and Hands-on Lab

  • API-based vs. on-prem model choices.
  • Building a retrieval-augmented agent with LangChain, FastAPI, and ChromaDB.
  • Simulated incident response and rollback procedures.

Summary and Next Steps

Requirements

  • A foundational understanding of Python programming.
  • Practical experience with data analysis or machine learning workflows.
  • Familiarity with healthcare data concepts (e.g., EHR, FHIR).

Audience

  • Healthcare data scientists and ML engineers.
  • Clinical informatics and digital health product teams.
  • IT leaders and innovation managers within the healthcare sector.
 14 Hours

Related Categories