CONF Burdisso_INTERSPEECH_2023/IDIAP Node-weighted Graph Convolutional Network for Depression Detection in Transcribed Clinical Interviews Burdisso, Sergio VILLATORO-TELLO, Esaú Madikeri, Srikanth Motlicek, Petr EXTERNAL https://publications.idiap.ch/attachments/papers/2023/Burdisso_INTERSPEECH_2023.pdf PUBLIC https://publications.idiap.ch/index.php/publications/showcite/Burdisso_Idiap-RR-03-2023 Related documents Proceedings of Interspeech 2023 We propose a simple approach for weighting self-connecting edges in a Graph Convolutional Network (GCN) and show its impact on depression detection from transcribed clinical interviews. To this end, we use a GCN for modeling non-consecutive and long-distance semantics to classify the transcriptions into depressed or control subjects. The proposed method aims to mitigate the limiting assumptions of locality and the equal importance of self-connections vs. edges to neighboring nodes in GCNs, while preserving attractive features such as low computational cost, data agnostic, and interpretability capabilities. We perform an exhaustive evaluation in two benchmark datasets. Results show that our approach consistently outperforms the vanilla GCN model as well as previously reported results, achieving an F1=0.84% on both datasets. Finally, a qualitative analysis illustrates the interpretability capabilities of the proposed approach and its alignment with previous findings in psychology. REPORT Burdisso_Idiap-RR-03-2023/IDIAP Node-weighted Graph Convolutional Network for Depression Detection in Transcribed Clinical Interviews Burdisso, Sergio VILLATORO-TELLO, Esaú Madikeri, Srikanth Motlicek, Petr depression detection Graph Neural Networks Interpretability limited training data node weighted graphs EXTERNAL https://publications.idiap.ch/attachments/reports/2023/Burdisso_Idiap-RR-03-2023.pdf PUBLIC Idiap-RR-03-2023 2023 Idiap June 2023 Paper submitted to INTERSPEECH 2023 In this work, we propose a novel approach for weighting self-connecting edges in a Graph Convolutional Network (GCN) and show its impact on depression detection from transcribed clinical interviews. To this end, we use a GCN for modeling non-consecutive and long-distance semantics to classify the transcriptions into depressed and control subjects. The proposed method aim to mitigate the limitation assumptions of locality and the equal importance of self-connections vs. edges to neighboring nodes in GCNs, while inheriting some attractive features such as, low computational cost, being data agnostic, and having interpretability capabilities. We perform an exhaustive evaluation in two benchmark datasets. Results show that our approach consistently outperforms the vanilla GCN model as well as previously reported results, achieving a SOTA F1=0.84% on both datasets. Finally, qualitative analysis shows an alignment between findings in psychology research and what the model learned from the data.