Unlocking Disease Insights: Data Mining and Precise Prediction of Kidney Disease in Visakhapatnam, India

Abstract

Data mining serves as an essential tool for comprehending datasets related to diseases. The data in question was collected within Visakhapatnam District of Andhra Pradesh, India, spanning the period from 2021 to 2022 and encompasses 1380 instances, equally divided into 690 instances of kidney disease and 690 instances of healthy subjects. This dataset leverages various health-related profiles, including age, height, weight, gender, blood pressure, blood sugar levels, water intake, and insulin levels, to forecast the likelihood of a patient developing kidney disease. Several methods, such as feed-forward neural networks, probabilistic neural networks (PNN) including confusion matrix analysis, unsupervised clustering via Self-Organizing Maps (SOM), and dynamic time series analysis for prospect prediction, were meticulously analyzed using the MATLAB platform. The outcomes indicate that each of these methods exhibits distinct strengths concerning specific data mining objectives. Notably, the dataset achieved a remarkable 100% accuracy in predicting kidney disease, underscoring its efficacy in this context.