2025 Volume 32 Issue 3 Pages 209-216
In recent years, research in mental health utilizing digital phenotyping has increased. In general, digital phenotypes are classified into two categories: active and passive data. Passive data include various biomarkers such as facial expression, heart rate, and walking speed. Among them, voice data have emerged as a significant biomarker.
Since 2003, a surge in research on emotion analysis using voice data has been observed. This trend has extended to mental health studies, and researchers worldwide have been actively exploring the use of machine learning and deep learning to diagnose, monitor, and treat mental health disorders.
This study presents a concise overview of the fundamental mechanisms of voice and emotion analysis and its application to mental health research. In the field of occupational health, digital mental health studies incorporating voice and emotion analysis are in the early stages. As such, significant potential remains to explore its role in mental health prevention. Lastly, this study discusses current challenges and future considerations in this evolving field.
