A Comparative Analysis of Clustering Algorithms in Targeted Preventive Care
In the modern healthcare, the importance of preventive care has came out as an effective approach in order to lighten the load of hospitals while also ensuring prolonged life span for patients. The current study on Clustering algorithms in targeted preventive care, highlights one such major issue in clustering algorithms that help uncover hidden patterns from a dataset to generate more precise preventative ac- tions in medical care. The core of this study involves a detailed comparative study of different clustering algorithms: K-Means, DBSCAN and Hierarchical Clustering. The methodology used in this study in- volves using various health datasets containing such features as age, body mass index, diabetes, asthma, arthritis, life style factors and geographical location, etc. The main evaluation factors to consider in regard to the algorithms include Silhouette scores metric, Scalability, Robustness and Interpretability. The results of this analysis, K-Means proves to be very efficient when dealing with huge, well-defined datasets. For this reason, it is suitable for population screening. However, when considering identification of outliers in patients having rare or unusual health problems, DBSCAN shows much better results. On the other hand, Hierarchical clustering gives much better insight into nesting various health risks. This report also focus on preprocessing, including scaling of data and handling incomplete values. Another interesting point of the analysis is that besides simple classification of health data, it may help predict chronic diseases, discovering some symptoms that would remain unnoticed during manual examination.
Authors
- Shalvi Tripathi
- Garima Singh
- Arushi Goyal
Publication Details
- Journal
- International Journal of Innovation and Multidisciplinary Research
- Published
- 2026-11-15
- DOI
- https://doi.org/10.67505/ijiamr.v6.i2.62001
- Primary Topic
- Advanced Clustering Algorithms Research
- Type
- article
- Field-Weighted Citation Impact
- 0.00