Machine Learning in the Service of Mental Health Management: Optimization Metaheuristics for Improving Disorder Classification

Mental health awareness has become an increasingly pressing societal issue in recent years. Many individuals struggle with undiagnosed disorders, leading to a significantly reduced quality of life. Consequently, timely diagnosis and effective treatment have become essential. Improving mental health care benefits not only individuals but also society as a whole by promoting overall well-being and productivity. However, limited resources and infrastructure often constrain patients' access to mental health professionals. This work seeks to explore the use of advanced machine learning-powered classification algorithms for detecting and identifying mental health disorders with higher accuracy. Since the performance of classification algorithms depends heavily on proper parameter selection, a modified metaheuristic optimization algorithm, based on the firefly algorithm, is introduced to enhance performance and reliability. The proposed approach was evaluated on a publicly available real-world dataset, and a detailed comparative analysis with several contemporary algorithms was conducted. The best-performing models achieved an accuracy exceeding 94%, suggesting the approach's strong viability for real-world assistive applications in mental health care.