The reliance on subjective self-reports in mental health assessment poses significant challenges regarding objectivity and timeliness.This study introduces an approach based on the intelligent multimodal analysis of musical time-series to mitigate these issues.Unlike conventional techniques that ignore the temporal dynamics of emotional response, our framework models the continuous interplay between music audio and physiological markers (such as heart rate) to decode evolving psychological states.We formulate a dynamic analysis mechanism that specifically targets the subtle patterns of emotion during music perception.Empirical results show significant improvements over static baselines, yielding an accuracy exceeding 85% and an AUC of 0.91 in detecting depressive tendencies.These findings validate the efficacy of analysing musical time-series as a robust, non-invasive method for mental health estimation.
Lv et al. (Thu,) studied this question.