ARTIFICIAL INTELLIGENCE (AI)-POWERED PREDICTIVE MODELING FOR PATIENT READMISSION AND TREATMENT RESPONSE USING ELECTRONIC HEALTH RECORDS AND MACHINE LEARNING

Abstract

Predictive modeling using electronic health records (EHRs) and machine learning can revolutionize the medical field by exposing high-risk patients and refining treatment strategies. Objective: Using EHRs and machine learning techniques, this research seeks to develop and evaluate AI-powered predictive models for patient readmission and treatment response. Methods: Using a quasi-experimental study design, the effectiveness of AI-powered predictive models in projecting patient readmission and treatment response was assessed. There were 200 adults, age (≥18 years), in the sample. Carried out in a hospital setting, the study uses electronic health records (EHRs) and allows evaluation of AI-driven predictive algorithms in a real clinical environment. Electronic health records (EHRs) are the primary data source for the study. EHRs give extensive data on patient demographics, treatment outcomes, and medical history. Descriptive statistics; logistic regression; machine learning algorithms (random forest, support vector machine); model performance evaluation using metrics such accuracy, precision, recall, and area under the receiver operating characteristic curve AUC-ROC. Results:  The model identified significant readmission risk factors with an 85% accuracy rate. Conclusions: By identifying high-risk individuals and fine-tuning treatment protocols, AI-powered predictive modeling has demonstrated its ability to improve patient outcomes.The findings suggest that clinical decision support systems providing personalized recommendations for patient care could be developed using artificial intelligence.