A forecasting approach for hospital bed capacity planning using machine learning and deep learning with application to public hospitals

Hospital Bed Capacity (HBC) planning affects economic and social sustainability in healthcare through bed capacity efficiency and medical treatment accessibility. Conventionally, this problem is solved using programming or simulation models with assumptions and limits. Forecasting the HBC using time series data on bed occupancy has been considered but not with factors such as the Number of Hospitalized Patients (NHP) and patient's length of stay (LOS). This study proposes a data-driven methodology to forecast the HBC using Machine Learning (ML) and Deep Learning (DL). The LOS classification is performed using several ML techniques, including the Bayesian network, K-nearest neighbor, support vector machine, decision tree, and Linear regression. Also, the seasonal autoregressive integrated moving average, linear regression and Long short-term memory neural network are applied for the NHP forecasting. The forecasting and descriptive analysis outputs based on LOS classes are directly applied to a simple mathematical model to predict the required bed capacity. This methodology is applied in a case study in a heart ward at a public hospital. The data set includes 51231 records, and ML and DL algorithms are developed in Python. Results show that the heart ward's bed capacity must be raised from 45 to 137 by 2026. In addition, several managerial recommendations are formulated.