Global Mental Health Disorder Trends: Advanced Forecasting Techniques Using Time Series and Neural Networks

Mental health disorders, such as schizophrenia, anxiety, depression, and substance abuse, have become critical global health concerns. This study aims to forecast the future prevalence of these disorders by applying both traditional statistical models and advanced deep learning techniques. Using three distinct datasets from Kaggle, we investigate the predictive capabilities of models including ARIMA, Holt-Winters Exponential Smoothing (HWES), Moving Average (MA), AutoRegressive (AR), Long Short-Term Memory (LSTM), and Recurrent Neural Networks (RNN). These models are evaluated across key performance metrics—MSE, MAE, RMSE, and SMAPE—to determine their effectiveness in capturing trends and forecasting mental health disorder prevalence. Our findings demonstrate that while deep learning models such as LSTM offer robust forecasting potential, HWES and ARIMA models exhibit superior accuracy for trend-based predictions, particularly in the case of schizophrenia and anxiety disorders. This research advances the field of mental health forecasting by providing comprehensive insights into the strengths and limitations of various predictive models, ultimately contributing to more effective mental health policy planning and intervention strategies on a global scale.