Depression-Anxiety Model
The strongest reported audio-only model reached 71.1% depression Sn = Sp and 70.7% anxiety Sn = Sp. The model scores are designed to relate monotonically to PHQ-9 and GAD-7 totals and map to clinical scale thresholds. The Depression-Anxiet…
1 sources - 5 claims
The strongest reported audio-only model reached 71.1% depression Sn = Sp and 70.7% anxiety Sn = Sp. The model scores are designed to relate monotonically to PHQ-9 and GAD-7 totals and map to clinical scale thresholds. The Depression-Anxiety Model estimates depression and anxiety severity scores from spoken audio. The study presents an audio-only screening model that works from at least 30 seconds of speech. The released HuggingFace model is intended for further research into speech-based mental health assessment.