ksb-d08-k0048

name: "KSB-D08-K0048" description: "Explainable AI (XAI) Methods for PV: Comprehensive understanding of explainable AI techniques applicable to pharmacovigilance including S..." version: "1.0.0" domain: "D08" domain_name: "Signal Detection & Management" type: "Knowledge" proficiency_level: "L4" bloom_level: "analyze" triggers: - "explain explainable ai (xai) methods for pv" - "what is explainable ai (xai) methods for pv" - "analyze explainable ai (xai) methods for pv" epa_mapping: "EPA-04, EPA-05" cpa_mapping: "CPA-02" regulatory_refs: ""

KSB-D08-K0048: Explainable AI (XAI) Methods for PV

Overview

Domain: D08 - Signal Detection & Management Type: Knowledge Proficiency Level: L4 (Proficient - Independent practice) Bloom Level: Analyze

Description

Comprehensive understanding of explainable AI techniques applicable to pharmacovigilance including SHAP (SHapley Additive exPlanations), LIME (Local Interpretable Model-agnostic Explanations), attention mechanisms, feature importance analysis, decision trees for model approximation, and counterfactual explanations. Knowledge of selecting appropriate XAI methods based on model type, stakeholder needs, and regulatory requirements per CIOMS WG XIV transparency principles.

Context

• Major Section: AI Integration and Advanced Analytics
• Section: AI Transparency and Explainability

EPA Mapping

• EPA-04:3009-3010
• EPA-05:3011-3012

CPA Pathway

• CPA-02

Regulatory References

• General PV competency

Instructions

When this skill is activated, Claude should:

1. Demonstrate L4 proficiency in explainable ai (xai) methods for pv
2. Apply analyze level cognitive skills to analyze the topic
3. Reference relevant regulatory guidance (general PV standards)
4. Connect to related EPAs: EPA-04, EPA-05

Key Competencies

• Comprehensive understanding of explainable AI techniques applicable to pharmacovigilance including SHAP (SHapley Additive exPlanations), LIME (Local Interpretable Model-agnostic Explanations), attention mechanisms, feature importance analysis, decision trees for model approximation, and counterfactual explanations. Knowledge of selecting appropriate XAI methods based on model type, stakeholder needs, and regulatory requirements per CIOMS WG XIV transparency principles.

Assessment Criteria

• Can analyze core concepts independently
• Demonstrates understanding of regulatory context
• Applies knowledge appropriately to PV scenarios

Related Skills

• Other D08 skills in AI Transparency and Explainability
• Cross-domain integrations per DAG architecture

Generated from PV KSB Framework v1.0 | 2025-12-31