Predictive Machine Learning Approaches for Mental Health Diagnoses in College Students

In recent years, the rise in mental health disorders among college students has spurred the adoption of machine learning for predictive diagnosis. This trend emphasizes the need for innovative strategies from educators, healthcare providers, and policymakers. The diagnostic process involves interviews, medical history assessments, and psychological tests, posing challenges and increasing the risk of misdiagnosis. Accurate diagnosis is crucial for college students, who face unique stressors such as academic pressures, social adjustments, financial constraints, and limited access to mental health resources, contributing to a disproportionately high prevalence of mental health issues in this demographic. Machine learning techniques, including Logistic Regression (LR), Random Forest (RF), Support Vector Machines (SVM), knearest Neighbors (KNN), and Convolutional Neural Networks (CNN), show promise in enhancing diagnostic accuracy and supporting mental health professionals in identifying conditions like Attention deficit /hyperactivity disorder (ADHD), Depression and anxiety, bipolar disorder, and Schizophrenia. This study focuses on preprocessing and analyzing a dataset to predict mental health issues using various machine learning approaches, highlighting Random Forest’s notable performance. Specifically, the Random Forest classifier achieved an accuracy of 0.85, along with a leading F1Score of 0.86 and Area under curve (AUC) of 0.95, while CNN classifiers achieved a perfect recall score of 1.0.