AI-POWERED PREDICTIVE ANALYTICS IN HEALTHCARE: ENHANCING DIAGNOSIS AND TREATMENT OUTCOMES

Abstract

The use of powerful analytics has helped AI in the healthcare industry to greatly improve diagnosis and care. The thesis discusses how AI models such as SVMS and CNNs are being used to strengthen disease prediction, treatment plans and the accuracy of diagnoses. Information was taken from credible medical sources; features were selected using PCA and correlation and every detail included was validated carefully. For the models used which include Random Forest, Logistic Regression and SVM, accuracy, precision, recall, F1-score and ROC-AUC were main factors used to compare and evaluate them. The model performed the best due to its 0.91 ROC-AUC and the accuracy rate of 86% over the other models. This work covers the benefits of making models understandable and usable within day-to-day medical care. Issues pertaining to generalizing models, the lack of training data and honouring individual privacy are being discussed. They clearly demonstrate that AI may improve healthcare through its support in decision-making for clinicians, better treatment choices and faster diagnosis. Future research in healthcare is focused on improving the visibility of AI data models, including various types of patients and making sure AI use is ethical.