Development of Data-Based Monitoring System for Management of Hospitalization Days in a Regional Tertiary Hospital of Korea.

As the use of medical services increases as we enter an aging society, the number of days of hospitalization is a major factor affecting medical expenses and needs to be managed. We aim to develop a data-based monitoring system for the management of hospitalization days in a regional tertiary hospital of Korea. The contents consisted of 1) development of a monitoring system for days of hospitalization, and 2) development of prediction model for days of hospitalization. The days of hospitalization were monitored and visualized automatically through an interactive R shiny web application. Based on the extracted variables, machine learning models were developed to predict patients hospitalized for more than 7 days, and the XGBoost model showed the best performance with the AUROC of 0.83. Efficient management of hospitalization days is expected by providing the appropriate monitoring system and model.