A Comparative Analysis of Unsupervised Learning Techniques in the Prediction of Cardiovascular and Liver Diseases

Early and accurate prediction of chronic diseases such as cardiovascular and liver disorders remains a critical challenge in healthcare. With the growing volume of medical data, unsupervised machine learning has emerged as a promising approach for uncovering hidden patterns in unlabelled datasets. This study conducts a comparative analysis of key unsupervised learning techniques, such as K-means clustering, DBSCAN, and hierarchical clustering, applied to medical datasets related to cardiovascular and liver conditions. The goal is to evaluate their effectiveness in grouping patients based on clinical attributes without relying on pre-labeled outcomes. The analysis focuses on clustering accuracy, silhouette scores, computational efficiency, and interpretability of results. Findings suggest that while K-means performs well on linearly separable data, DBSCAN offers robustness against noise and varying densities, making it more suitable for complex medical data. Hierarchical clustering, on the other hand, provides better visual insight into patient subgroup structures. The results underscore the importance of selecting appropriate algorithms based on data characteristics and diagnostic goals. This research contributes to the advancement of data-driven disease prediction and highlights the potential of unsupervised learning in personalized healthcare.