This research focuses on the development and evaluation of a semi-supervised clustering model aimed at improving the accuracy and effectiveness of patient data clustering in smart, person-centric healthcare systems. By integrating both supervised and unsupervised learning methods, the proposed model seeks to address the limitations of traditional clustering algorithms like K-means, Hierarchical Clustering, and DBSCAN, especially in handling high-dimensional, heterogeneous healthcare data that reflects individual patient characteristics. Extensive experiments were conducted on a dataset of 132,458 patient records, with a focus on metrics such as Silhouette Score, Davies-Bouldin Index (DBI), Inertia, Adjusted Rand Index (ARI), and Normalized Mutual Information (NMI). Results demonstrate that the proposed model outperforms baseline algorithms across all metrics, showing significant improvements in cluster cohesion, separation, and alignment with true labels. The findings suggest that integrating semi-supervised learning with clustering methods provides a more robust framework for supporting personalized treatment, enhancing disease prediction, and optimizing healthcare resources to align with individual patient needs in a truly person-centric approach.