Machine Learning-based Decision Support System for Healthcare in the Context of COVID-19: Case Study of Saudi Arabia

Abstract

The COVID-19 pandemic has highlighted the significance of effective healthcare decision-making and resource allocation, posing substantial problems for healthcare systems worldwide, including Saudi Arabia. This work describes the creation of a Machine Learning-based Decision Support System (DSS) designed to improve the management of COVID-19 outcomes, with a focus on hospitalization, recovery, and death rates. The study uses a variety of machine learning methods, including Decision Trees, Linear Regression, Random Forest, and SARIMAX, to examine large datasets of daily confirmed cases, recoveries and fatalities. The Random Forest model outperformed the Linear Regression and SARIMAX models in terms of predicted accuracy. Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) figures demonstrate the Random Forest model's superiority, which is particularly good in reflecting the complexity of COVID-19 spread. Furthermore, the study underlines the value of extensive datasets, good feature selection, and model validation in improving prediction accuracy. These findings have significant implications for healthcare practitioners and policymakers, allowing for more informed decision-making and effective resource management during the current pandemic. As a result, the study calls for continuous model refining, multidisciplinary collaboration, and real-time data integration to increase the impact of machine learning applications in healthcare, ultimately leading to better patient outcomes and responsiveness during public health crises.