Explainable counterfactual reasoning in depression medication selection at multi-levels (personalized and population)

This study investigates how variations in Major Depressive Disorder (MDD) symptoms (HAM-D) are associated in a predictive model with randomized clinical trial (RCT) arm assignment between SSRIs and SNRIs. We applied explainable counterfactual reasoning with counterfactual explanations (CFs) to assess the impact of specific symptom changes on model-predicted RCT arm assignment. Across 17 classifiers, CatBoost achieved the highest performance; typical test metrics ranged 0.74–0.78 with best ROC-AUC 0.7640. Sample-based CFs revealed both local and global feature importance of individual symptoms in medication selection. Counterfactual reasoning highlights which MDD symptoms the model uses to distinguish SSRI vs. SNRI trial assignments, supporting interpretable AI-based decision support while requiring prospective real-world validation beyond the RCT context. Future work should validate these findings on more diverse cohorts and refine algorithms for clinical deployment.