Exploring the Efficiency of Advanced ML Techniques for Multi-Dimensional Cardiovascular Disease Prediction and Prognostic Analysis

Since cardiovascular diseases (CVDs) remain a main cause of death globally, accurate and efficientdiagnosis tools are necessary. This work explores the use of machine learning methods, namely K-NearestNeighbors (KNN) and Random Forest (RF), for the prediction of cardiovascular diseases based on theCleveland Heart Disease dataset. Including age, blood pressure, cholesterol, and ECG measurements, thedataset comprises 14 continuous and categorical traits covering 7,000 individuals. Many data preparationmethods feature scaling, data splitting, one-hot encoding, and handling missing values were applied toensure the dataset was fit for model training. The predictive power of both Random Forest and KNN wasselected; Random Forest is an ensemble method combining forecasts from numerous decision trees, whilstKNN is a non-parametric classification algorithm based on the majority vote of its nearest neighbors.Hyperparameter adjustment for both models using grid search and cross-validation helped to identifyoptimal values for model performance. The models were evaluated using key performance measurescomprising accuracy, sensitivity, specificity, and Area Under the ROC Curve (AUC).Random Forest performed better than KNN in all evaluated criteria: its 92% accuracy, 89% sensitivity,94% specificity, and 0.95 AUC show this. Conversely, KNN boasts an accuracy of 86% but a reducedsensitivity and specificity. The results show that since Random Forest is better at separating patients withcardiovascular illness from those without, it offers a potential tool for early diagnosis. The results of thestudy indicate that by increasing early identification of cardiovascular diseases, machine learning modelsRandom Forest have enormous potential to improve patient outcomes and save healthcare expenses.