Diabetic Prediction using Biased Renovate K Means Clustering

Diabetes Mellitus has arisen as a major global healthcare issue, impacting millions of people globally. Diabetes early identification and prediction are crucial for optimal prevention and treatment. Machine learning approaches have been critical in diabetes prediction, with K-Means clustering being a key unsupervised learning algorithm. Traditional K-Means clustering, on the other hand, is fundamentally vulnerable to initialization and might result in poor cluster assignments, thus leading to incorrect diabetes predictions. In order to enhance the precision and consistency of diabetes prognosis, this research work proposes a novel method named “Biased Renovate K-Means clustering” (BRKMC). The classic K-Means method is combined with a biasing mechanism that changes cluster initiation depending on pertinent medical and demographic data in BRKMC. This biasing guarantees that the clustering procedure is more sensitive to crucial diabetes risk indicators, enhancing the identification of high-risk people. The BRKMC algorithm is applied to a large dataset that includes medical records, lifestyle data, and genetic markers from a varied population. BRKMC successfully identifies subpopulations with a greater vulnerability to diabetes by improving the initialization of cluster centroids. Following that, these subpopulations are examined, and predictive models are created to assess the risk of diabetes for people inside each cluster.