Predicting patient visits at the psychiatric polyclinic of a public hospital in Bali, Indonesia: A forecasting approach using single exponential smoothing

Background: Hospital planning requires effective management of resources, facilities, and costs, and accurate patient visit forecasting is integral to this process. Forecasting methods, such as single exponential smoothing, are widely used to predict patient visits and aid in resource allocation. However, forecasting for psychiatric polyclinics presents unique challenges due to the fluctuating nature of mental health conditions and the difficulty in predicting patient behavior outside clinical settings. Objective: This study aimed to apply the single exponential smoothing method to forecast psychiatric polyclinic visits at a public hospital in Bali province, Indonesia, for the years 2024–2026, with the goal of assisting hospital management in planning resources and services. Methods: A retrospective analysis was conducted using patient visit data from January 2021 to December 2023. The data was processed and analyzed using the single exponential smoothing method through SPSS software. Forecasting accuracy was assessed using Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE), with forecasts for 2024–2026 generated for monthly patient visits. Results: The forecast for psychiatric polyclinic visits predicted a steady increase in visits, with 9,320 visits in 2024, 9,627 visits in 2025, and 9,939 visits in 2026. The accuracy of the forecasting model was confirmed by a MAPE of 2.87%, indicating high precision. The RMSE value was 26.8%, suggesting the average forecasting error was 26.8%. Conclusion: The single exponential smoothing method proved to be a reliable and straightforward tool for forecasting psychiatric polyclinic visits. With a high degree of forecasting accuracy, the method offers practical applications for hospital administrators to plan for future patient volume, optimize resource allocation, and ensure sufficient service capacity. Future research could explore the use of additional forecasting models and broader datasets to enhance predictive accuracy.