Artificial Intelligence in Healthcare: Predictive Analytics for Early Detection of Chronic Diseases

Abstract

The importance of early diagnosis of chronic diseases will help in minimizing disease burden, reduce the outcome of patients, and decrease the cost of healthcare in the long-term. New technologies in the field of Artificial Intelligence (AI) allow applying predictive analytics to diagnose disease risk at an early stage using big data on clinical analysis. In this research, the researchers examine the capability of AI-based predictive models to identify early chronic disease through anonymized electronic health record data. An analysis of 30,214 patient records with demographic variables, clinical variables, laboratory variables, lifestyle variables, and medical history variables was performed. Five machine learning models applying 18 key predictors to develop and compare were created on the post-processing and feature selection: Logistic Regression, Decision Tree, Random Forest, Support Vector Machine and Deep Neural Network (DNN). The performance of the model was measured in terms of accuracy, precision, recall, and F1-score, and AUC-ROC, and the focus was on recall and AUC-ROC as these indicators are necessary in early screening. The findings indicate that the improved AI models are more superior to conventional models, and the DNN attains the highest recall (0.91) and AUC-ROC (0.94), and then the Random Forest model comes next, with the recall = 0.90, and AUC-ROC = 0.93. The analysis of feature importance showed that the most influential predictors were age, level of fasting glucose, body mass index, blood pressure (systolic), and family history, which is consistent with the existing clinical risk factors. In general, the results show that AI-based predictive analytics can be an efficient and trustworthy clinical decision-support second-wave tool to predict chronic illnesses in their early stages, which promotes the transition to preventive, data-driven healthcare.