Modelling the length of hospital stay after knee replacement surgery through Machine Learning and Multiple Linear Regression at “San Giovanni di Dio e Ruggi d'Aragona” University Hospital

Abstract

Knee arthroplasty is one of the most commonly performed procedures within a hospital. The progressive aging of the population and the spread of clinical conditions such as obesity will lead to an increasing use of this procedure. Therefore, being able to make the process related to this procedure more effective and efficient becomes strategic within hospitals, subject to increasingly stringent clinical and financial pressures. A useful parameter for this purpose is the length of stay (LOS), whose early prediction allows for better bed management and resource allocation, models patient expectations and facilitates discharge planning. In this work, the data of 124 patients who underwent knee surgery in the two-year period 2019-2020 at the San Giovanni di Dio and Ruggi d'Aragona university hospital were studied using multiple linear regression and machine learning algorithms in order to evaluate and predict how patient data affect LOS.