Deep Learning-Based Predictive Modeling for Male Depression Detection

Abstract

This project utilizes machine learning techniques to construct a highly precise model for categorizing audio recordings, with a particular focus on male speakers and their mental health conditions. The audio recordings are classified into three distinct categories: Remitted (RMT), Depressed (DPR), and High-risk for suicide (HRK), with special attention to gender-specific nuances. We have conducted an extensive exploration and comparison of diverse machine learning models, including 1D and 2D Convolutional Neural Networks (CNNs), Support Vector Machine (SVM), and Recurrent Neural Network with Long Short-Term Memory (RNN-LSTM). Our primary goal is to identify the most accurate model for classifying these male audio recordings, potentially offering a valuable tool for early detection and intervention in male mental health issues. We eagerly look forward to sharing our research results, aiming to make a substantial contribution to the understanding and treatment of depression among males. In this paper, we present the results of our investigation, comparing the accuracy of audio classification using 25-second and 1-minute speech segmentation.