A Holt-Winter Approach to Forecasting Admission in the Psychiatric Department

Psychiatric admissions predictive modeling is a crucial step in hospital resource and service management. Prediction models are essential for medical facilities to account for patient arrival fluctuations that could potentially overextend their resources. The present study used the Holt-Winters method to forecast psychiatric admissions at Hospital Universiti Sains Malaysia (HUSM). The estimation and evaluation phases are based on historical admission data from 2017 to 2024 to extract trends, seasonal and level components. The dataset was split into training and testing phases to ensure robust model selection. Mean Absolute Scaled Error (MASE), Root Mean Squared Error (RMSE) and Mean Absolute Percent Error (MAPE) are used to compare the performance. This paper examines two forms of the Holt-Winters model: additive and multiplicative. The results reveal that the Additive Holt-Winters model consistently has the lowest error across all performance metrics. Thus, it is the best for predicting psychiatric patient admissions. The forecasting results indicate an overall increase in admissions with variations driven by seasonal factors. An accuracy assessment reveals that most predictions match actual admissions to some extent, suggesting the model is reliable. These include certain months that demonstrate standout differences, suggesting that other confounding factors may be influencing admission rates. It demonstrates the robustness of Holt-Winters’ forecasting methodology in healthcare and its implications for hospital administrators seeking to improve efficiency in the deployment of resources. Future work might also focus on combining other predictive tools to improve forecast quality.