Exploring Algorithmic Explainability: Generating Explainable AI Insights for Personalized Clinical Decision Support Focused on Cannabis Intoxication in Young Adults

Explainable AI can make clinical decision-making more transparent: The study demonstrates that techniques like SHAP and decision trees can break down how AI systems reach conclusions about cannabis intoxication, allowing clinicians to understand which specific factors (such as heart rate or environmental noise) are driving recommendations rather than treating the AI as a "black box."

Individual differences in drug response require personalized approaches: The research found significant variation in how different young adults respond to cannabis use, including differences in activity patterns, sleep disruption, and timing of use, suggesting that one-size-fits-all clinical interventions may be ineffective and personalized analysis is necessary.

Multiple explainability techniques provide complementary insights: Rather than relying on a single explanation method, the study combined four different XAI approaches (SHAP, SkopeRules, decision trees, and counterfactual models), each revealing different aspects of cannabis use patterns—demonstrating that responsible AI implementation may require multiple analytical lenses.

Explainable AI could improve clinical trust and intervention design: By helping clinicians understand the reasoning behind AI-generated insights about patient behavior and physiology, XAI technologies could enable more targeted, evidence-based intervention strategies tailored to individual patient needs rather than generic recommendations.