Text-based depression detection on sparse data

TL;DR

This paper introduces a multi-task BGRU network with pretrained embeddings for depression detection in sparse clinical conversation data, demonstrating improved performance over existing methods.

Contribution

It proposes a novel multi-task loss function and investigates the effectiveness of sentence-level embeddings and pretraining for depression detection in sparse data scenarios.

Findings

Pretraining improves depression detection accuracy.

Sentence-level embeddings outperform word-level embeddings.

Mean and attention pooling are more effective than last-timestep pooling.

Abstract

Previous text-based depression detection is commonly based on large user-generated data. Sparse scenarios like clinical conversations are less investigated. This work proposes a text-based multi-task BGRU network with pretrained word embeddings to model patients' responses during clinical interviews. Our main approach uses a novel multi-task loss function, aiming at modeling both depression severity and binary health state. We independently investigate word- and sentence-level word-embeddings as well as the use of large-data pretraining for depression detection. To strengthen our findings, we report mean-averaged results for a multitude of independent runs on sparse data. First, we show that pretraining is helpful for word-level text-based depression detection. Second, our results demonstrate that sentence-level word-embeddings should be mostly preferred over word-level ones. While the choice of pooling function is less crucial, mean and attention pooling should be preferred over last-timestep pooling. Our method outputs depression presence results as well as predicted severity score, culminating a macro F1 score of 0.84 and MAE of 3.48 on the DAIC-WOZ development set.