Evaluating the Impact of Machine Learning Models in Forecasting Patient Outcomes: An EHR-Based Study

Abstract

Machine learning (ML) as a method of procedures has transformed healthcare especially in the outcomes forecasting process of patients. EHRs contain huge volumes of patient-related information and hence, possible opportunity exists to administer ML models to perform predictive analysis. The study is concerned with the application of machine learning to forecast the destiny of patients in accordance with the EHR data. The analysis will involve a comparison of various algorithms, i.e., Random Forest (RF), Support Vector Machine (SVM), and Neural Networks (NN) in order to determine the fate of the patients in terms of disease advance and risk of readmission. The tactics that were used in the current investigation may be listed as feature selection, data preprocessing, application of the ML algorithms to a publicly available EHR dataset. The accuracy of performance measures is based on the performance of the models, precision recall, and F1-score, upon which they are tested. Based on the findings, although neural models show that this neural model gives high accuracy rates, it is observed that they are likely to perform better in clinical use applications compared to Random Forest models due to their probability of interpretability. The paper helps in the advancement of knowledge on the application of machine learning in healthcare which would culminate in better patient care and decision making. In the future, more advanced methods of deep learning are to be applied and the transparency of the model is to be enhanced.