Efficient Machine Learning Tree-Based Models for Recognition of Chronic Diseases Using Big Data Health Records

Abstract

Today's people suffer from a wide variety of diseases due to various influences and choices made at the community level. Thus, to prevent the occurrence of such illnesses, persistent identification and prediction are paramount. Manually determining the disorders is generally challenging for doctors to be accurate with the exact numbers. Using massive data extracted from EHRs, this research lays forth an effective machine learning (ML) approach for CKD early diagnosis. Data preparation steps (including outlier removal, missing value replacement and transforming categorical data) are done before using normalization and RFE to find the best features. ETC is used as the main classification model because it helps to improve prediction and reduces the chances of overfitting by splitting the data randomly. With an accuracy of 99.5%, the model is very effective in diagnosing CKD. When evaluation measures include precision, recall, F1-score, and AUC-ROC, it shows that the approach performs well. They prove that using machine learning and big data together can enhance how early diagnosis and decisions are made in chronic disease cases.