A Novel Adaptive Machine Learning Model for Voice-Based Depression Detection with Performance Optimization

DOI

10.2991/978-94-6239-789-7_44How to use a DOI?

Keywords

Depression Detection; Machine Learning; Threshold Optimization

Abstract

In recent times, depression has become a growing global health concern. However, its early detection through non-invasive methods, such as speech/voice analysis, has achieved widespread popularity. This work represents an adaptive machine learning model for voice-based depression detection. We have compared the performance of the model with conventional approaches such as SVM and basic Decision tree methods. The proposed Adaptive Decision Tree (ADT) model outperforms the conventional DT model by leveraging dynamic feature selection, optimal threshold tuning, and ROC-based performance optimization. The adaptive threshold optimization approach improves recall in depression detection. Contrary to conventional fixed-threshold approaches, the proposed method vigorously adjusts the decision boundaries and minimizes false detection cases. It makes the method suitable for early screening of depression in a person. Furthermore, class imbalance was handled with pruning strategies to reduce overfitting and improve the generalization of the method. These developments make the model more suitable for clinical depression detection tasks by achieving a better balance between sensitivity and precision Experimental results reveal that the proposed model achieves superior performance, with an accuracy of 96.4%, precision of 95.5%, recall of 95.3%, F1 score of 95.6%, and AUC of 0.952. The adaptive SVM achieves high recall (96.8%) at the expense of precision. The proposed ADT model provides a balance between precision and sensitivity. It makes it a reliable and suitable approach for detecting clinical depression.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Kirti Bhatia
AU  - Renuka Arora
PY  - 2026
DA  - 2026/10/08
TI  - A Novel Adaptive Machine Learning Model for Voice-Based Depression Detection with Performance Optimization
BT  - Proceedings of theInternational Conference on Innovation and Practices in Artificial Intelligence (ICIPAI 2026)
PB  - Atlantis Press
SP  - 666
EP  - 679
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-789-7_44
DO  - 10.2991/978-94-6239-789-7_44
ID  - Bhatia2026
ER  -