IDENTIFYING INFLUENTIAL FACTORS BEHIND PHYSICIAN PERFORMANCE: A MACHINE LEARNING APPROACH TOSUPPORT DECISION-MAKING

Performance improvement is essential to every organization including medicine. For Human Resource to evaluate levels of performance quality, standards should be established as a base of developing evaluation methods. This paper aims to facilitate and enable more efficient performance evaluation and improvement methods for physicians by discovering the most reliable features. It will identify the main features influencing physicians' performance including line managers, patients, and physician information. These features are represented in a use case design that illustrates performance appraisal behavior. The paper presents a data set that includes 500 physicians in an Egyptian hospital, which was created based on the proposed features, and features affecting the physician’s performance will be tested using Regression analysis. Both the Lasso and Ridge regression models showed significant performance improvements. Lasso achieved an R² of 0.413 and an RMSE of around 0.01264, while Ridge achieved a slightly lower R² of 0.397 and an RMSE of around 0. 01281. The improved performance of these models can be attributed to their incorporation of regularization terms. Finaly, if these features are extracted automatically, they will better support the decision-making process. These features may help design a framework that helps evaluate physician performance and supports decision-makers in the future.