- Research Article
- 10.47164/ijngc.v16i3.1939
A Machine Learning-Based Multi-Label Classifier for Mental Health Condition Prediction Using Lifestyle and Emotional Behavior Data
- Mar 17, 2026
- INTERNATIONAL JOURNAL OF NEXT-GENERATION COMPUTING
- Kaustav Sanyal + 4 more +4
The increased rate of mental illness conditions like depression, anxiety, insomnia, and stress has led to the pursuit of computational techniques for early intervention and detection. The work in this paper introduces a multi-label classification model with machine learning that estimates multiple mental health conditions at once based on lifestyle and emotional behavior. A first-hand data collection was done using a survey that was specially designed, where data on sleep habits, mood, diet, exercise, emotional support received, and stress coping mechanisms was gathered. Labels were constructed that denote the existence of certain mental illnesses from answers to behaviorally suggestive questions. Standard preprocessing including imputation, normalization, and one-hot encoding was done to the data. A Random Forest classifier, packaged in a One-vs-Rest environment, was trained to carry out multi-label classification. The model was assessed on accuracy, precision, recall, and F1-score for every mental health condition. The outcomes show that lifestyle and emotional markers can be strong predictors, and the model is capable of achieving good classification performance on all target conditions. This strategy points out the possibility of machine learning in enhancing mental health monitoring and screening, particularly in non-clinical environments.
Read more