The feasibility of developing a neural network to detect signs of depression from speech samples using machine learning was evaluated in this cross-sectional study. Participants provided voice responses to the prompt “how was your day?” and self-reported PHQ-9 scores.
The results scientifically validate that machine learning voice biomarker technology can reliably detect signs and symptoms of depression from less than a minute of free form speech. Harnessing this technology with practitioners can expedite and prioritize patients for screening to help close the gap in depression detection.
This study was developed in partnership with UAMS and presented at the AAEP NUBE conference.
Click here to read the full abstract.