Machine learning (ML) has revolutionized predictive healthcare by enhancing early detection, diagnosis, and management of chronic diseases. This study presents a comparative analysis of ML algorithms for diabetes management and breast cancer detection. The research evaluates the effectiveness of Random Forest, Logistic Regression, Support Vector Machines (SVM), Decision Trees, and Linear Regression in predicting patient outcomes. The diabetes dataset includes medical and demographic factors such as age, BMI, HbA1c levels, and glucose levels. In contrast, the breast cancer dataset comprises tumor-related attributes like clump thickness, uniformity of cell size, and marginal adhesion. The results indicate that Random Forest consistently achieves the highest accuracy across both use cases, demonstrating its robustness in handling complex medical datasets. For diabetes prediction, Random Forest outperformed other models with an accuracy of 90.78%, while breast cancer detection achieved a classification accuracy of 96.50%. Logistic Regression and SVM also showed promising results but were less effective in handling non-linear relationships and high-dimensional data. While interpretable, decision Trees and Linear Regression required more extensive datasets to achieve comparable accuracy. This research highlights the potential of machine learning (ML) to enhance public health and lower healthcare costs through early diagnosis and personalized treatment. By integrating predictive models into clinical workflows like Electronic Health Records (EHRs), timely interventions and better resource allocation can be achieved, improving patient outcomes.