Pharmacovigilance Adverse Event Signal Triage Workflow Plan
Architect an automated multi-agent triage chain for adverse event intake, MedDRA coding, and regulatory deadline escalation.
Use this template when engineering an end-to-end drug safety processing pipeline that ingests adverse event data from multiple channels. It assists pharmacovigilance operations directors in building compliant, auditable multi-agent intake workflows.
Role: Director of Safety Systems and Regulatory Pharmacovigilance Engineering.
Context
- Target Safety Database: {{safety_database_type}}
- Source Intake Streams: {{intake_data_channels}}
- Terminology Standard: {{meddra_version}}
- Compliance Timeline SLA: {{reporting_timeline_sla}}
- Automated Triage Confidence Threshold: {{confidence_threshold_score}}
- Audit Requirement: {{audit_trail_requirements}}
Task
Produce an operational systems report detailing a multi-agent workflow chain that ingests multi-channel adverse event narratives, validates minimum safety criteria, maps terms using {{meddra_version}}, and routes prioritized cases to {{safety_database_type}} within {{reporting_timeline_sla}}.
Method
- Define the ingestion agent architecture capable of standardizing text across {{intake_data_channels}}.
- Design the ICSR Four-Element verification agent (identifiable patient, identifiable reporter, suspect product, adverse event).
- Detail the semantic extraction and MedDRA coding agent for mapping narrative symptoms against {{meddra_version}} preferred terms.
- Specify the seriousness evaluation agent to flag expedited fatal, life-threatening, or disabling events.
- Construct the dynamic routing agent that directs cases meeting {{confidence_threshold_score}} directly to draft case creation in {{safety_database_type}}.
- Formulate fallback mechanisms for low-confidence or ambiguous events requiring immediate safety specialist triage.
- Establish timing monitors to ensure safety cases do not breach the mandated {{reporting_timeline_sla}}.
- Specify the logging agent configuration to satisfy {{audit_trail_requirements}} across all agent transitions.
Constraints
- MUST flag potential 7-day and 15-day expedited reports within the first execution step.
- MUST NOT commit fully automated case submissions without human validation if confidence falls below {{confidence_threshold_score}}.
- Intermediate agent reasoning states MUST remain fully auditable under {{audit_trail_requirements}}.
- Architecture must support simultaneous ingest from all defined {{intake_data_channels}}.
Output format
Present a technical workflow plan structured into:
- Ingestion Pipeline & Channel Normalization
- Four-Criteria Validation & MedDRA Agent Logic
- Triage Routing, SLA Monitoring, & Escalation
- Compliance Architecture & Verification Log Limit the document to 800-1,200 words.
Self-review
- Does the workflow cover ICSR four-element validation comprehensively?
- Is the threshold {{confidence_threshold_score}} clearly applied as a decision boundary?
- Are strict timelines enforced for {{reporting_timeline_sla}} alerts?
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.