Item-Specific Similarity Assessments for Explainable Depression Screening

Depression, a prevalent mental health issue worldwide, is deeply influenced by the cultural and sociodemographic context, particularly in the Yucat ' an region in Mexico. Traditional depression screening methods, relying on self-reported questionnaires, often fall short in capturing the patterns of depressive symptoms specific to the area. This study introduces an innovative case-based reasoning (CBR) approach for depression screening, utilizing an item-specific similarity measure optimized through genetic algorithms. The main goal is to demonstrate the benefits of optimized similarity metrics to offer personalized reasoning capabilities that account for individual differences, thereby overcoming some of the limitations of one-size-fits-all retrieval approaches and achieving an accuracy comparable to other state-of-the-art machine learning (ML) alternatives. In contrast to these ML models, the proposed CBR approach has the additional benefit of being inherently explainable. Understanding the similarity between individual response patterns and cases of depression symptomatology allows for a more specific and culturally sensitive screening process. The findings suggest that this method could improve early detection and intervention strategies, offering an alternative to traditional scoring of self-reported depression questionnaires as well as a path toward more timely and effective mental health care in regions with unique cultural and demographic characteristics.