DATA SCIENCE FOR HEALTHCARE: MACHINE LEARNING APPLICATIONS IN PREDICTING PATIENT OUTCOMES

Machine learning (ML) has established itself as an invaluable tool in healthcare, providing advanced predictive capability across diverse patient outcomes. Different machine learning models from logistic regression, decision trees, support vector machines (SVM), random forests, gradient boosting machines (GBM), and convolutional neural network (CNN) have been applied in this study to identify some life-threatening patient end-point events including disease progression, emergency room triage, cancer prognosis and personalized treatment plans. Finally these models are evaluated for performance with respect to accuracy, computational cost and interpretability. It illustrates the pros and cons of each model in real healthcare applications and also describes the problems of patient privacy, model interpretability, and ethics. Machine learning models often aid in increasing diagnostic accuracy, decreasing the cost of healthcare, and improving the individualization of health care. The results indicate that even though complex models like CNN and SVM exhibit higher accuracy, they are overkill and expensive computationally and less interpretable in practical clinical settings. This study adds to the literature on the use of machine learning in health care and describes how predictive models may provide guidance regarding how to provide better patient care and patient outcomes.