Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 14 hours
Course Outline
Fundamental Principles of Agentic AI in Healthcare
- Distinguishing agentic systems from standard tool-based LLM applications.
- Defining autonomy limits, operational policies, and the role of human oversight.
- Navigating the healthcare data environment and associated constraints (EHR, FHIR, PHI).
Architecting Agent Workflows
- Implementing planning, memory, tool utilisation, and reflective cycles.
- Applying prompt engineering, function/tools, and strategic action selection.
- Managing state and adopting effective orchestration patterns.
Retrieval-Augmented Agent Development
- Ingesting and segmenting medical documentation for optimal processing.
- Utilising embeddings, vector databases, and assessing relevance.
- Ensuring response accuracy and implementing citation methodologies.
Healthcare System Integration and Interoperability
- Introduction to FHIR and SMART standards for agent connectivity.
- Processing both structured and unstructured clinical data effectively.
- Managing event-driven architectures, APIs, and comprehensive audit trails.
Safety, Risk Management, and Governance
- Establishing guardrails, conducting red-teaming, and designing fail-safes.
- Handling PHI, executing de-identification, and enforcing access controls.
- Implementing human-in-the-loop reviews and defining escalation protocols.
Evaluation and Continuous Monitoring
- Conducting offline assessments, defining golden datasets, and setting KPIs.
- Detecting hallucinations and performing rigorous factuality verification.
- Enhancing observability, logging, and managing cost and latency.
Deployment Strategies and Practical Laboratory Session
- Comparing API-based solutions versus on-premises model deployments.
- Constructing a retrieval-augmented agent using LangChain, FastAPI, and ChromaDB.
- Simulating incident response scenarios and practising rollback procedures.
Conclusion and Future Pathways
Requirements
- Proficiency in fundamental Python programming concepts.
- Practical experience with data analysis or machine learning pipelines.
- Awareness of healthcare data standards and structures (e.g., EHR, FHIR).
Target Audience
- Healthcare data scientists and machine learning engineers.
- Clinical informatics specialists and digital health product development teams.
- IT executives and innovation leads within the healthcare sector.
