Depression and Anxiety Prediction Using Deep Language Models and Transfer Learning

TL;DR

This study demonstrates that deep language models combined with transfer learning can effectively predict depression and anxiety from conversational speech, achieving high accuracy and revealing distinct linguistic cues for each condition.

Contribution

The paper introduces a novel approach using deep language models and transfer learning to detect depression and anxiety from speech data, with detailed analysis of linguistic cues.

Findings

Binary classification AUC ranges from 0.86 to 0.79

Best results occur when co-occurrence is either both or neither condition

Word sequence cues are more salient for depression than for anxiety

Abstract

Digital screening and monitoring applications can aid providers in the management of behavioral health conditions. We explore deep language models for detecting depression, anxiety, and their co-occurrence from conversational speech collected during 16k user interactions with an application. Labels come from PHQ-8 and GAD-7 results also collected by the application. We find that results for binary classification range from 0.86 to 0.79 AUC, depending on condition and co-occurrence. Best performance is achieved when a user has either both or neither condition, and we show that this result is not attributable to data skew. Finally, we find evidence suggesting that underlying word sequence cues may be more salient for depression than for anxiety.