[18] MACHINE LEARNING BASED DECISION SUPPORT SYSTEM ON ELECTRONIC HEALTH RECORD DATA TO PREDICT HEALTH SCORE

How to Cite the Article: Pramit Kumar Samant, Vinay Pathak & Waqar Ahmad (2026). Machine Learning Based Decision Support System on Electronic Health Record Data to Predict Health Score. International Journal of Multidisciplinary Research & Reviews, 5(7),239-274. https://doi.org/10.56815/ijmrr.v5i7.2026.239-00

Abstract

Machine Learning as well as AI developed systems are now today greatly used in varieties of real-life applications. E-healthcare management system is also one of them. The decision support system developed using many machine learning algorithms provides to optimize decision in protective health care systems. The proposed system consists of two modules. The first module included a novel Electronic Health Record based decision support system using SVM technology structure. The second module be composed of preprocessing, data extraction, exploratory data analysis from EHR data set. The proposed EHR based decision system is tested with accuracy for detecting health status on 79540 records. EHR data are mostly important for evaluating or predicting the health status after observing the varieties of medical parameters like body temperature, heart rate, blood pressure in both range systolic + diastolic and oxygen saturation level etc. How the new sample in the group data analysis can be achieved through machine learning approach and how this approach can be used as in AI tool for E-healthcare system. Also, visualization of data (exploratory data analysis i.e., EDA) is important to clarify about relationship among the parameters how they are closed to each other to evaluate the actual prediction of health score. In our data analysis, we have shown Support vector machine i.e., SVM have the highest accuracy rate (near about to 93%) among the other chosen classifiers as well as simple neural networking approach. The result follows on accuracy, recall, precision and confusion matrix are also shown in this evaluation. Violin plot and two-dimensional kernel density estimation (2D KDE) plots are also implemented for visualization of the dataset.