Major depressive disorder (MDD) and anxiety are increasingly prevalent among university student populations, yet early detection remains reliant on psychometric instruments tied to diagnostic criteria (e.g., PHQ-9, GAD). Emerging evidence suggests that depression affects both the acoustic properties and content of speech, making speech analysis a promising candidate as a digital biomarker for early screening.
This study evaluates the validity of acceXible, a speech-based machine learning platform, for the detection and monitoring of depression and anxiety in the student population of the Universidad Autónoma de Coahuila (UAdeC), Mexico. AcceXible captures spontaneous speech through open-ended interview tasks and applies automated acoustic and linguistic analysis.
The primary objective is to evaluate the validity of the acceXible spontaneous speech analysis system for depression and anxiety screening, assessed against the PHQ and GAD scales as reference standards. Secondary objectives include examining associations between speech-derived variables and other study measures, evaluating participant engagement with digital mental health resources, assessing user satisfaction with the platform, and analyzing longitudinal changes in scores across follow-up assessments.