A Comprehensive Analysis of KNN Variants for Disease Classification: Performance Evaluation and Future Directions

Accurate disease classification is vital for effective diagnosis and treatment planning. K-Nearest Neighbor (KNN) is a widely used supervised machine learning algorithm due to its simplicity and reliability. This paper presents a comprehensive comparative analysis of various KNN variants applied to disease classification. The study evaluates Classic KNN, Adaptive KNN, Locally Adaptive KNN, Fuzzy KNN, K-means Clustering KNN, Weight Adjusted KNN, Hassanat KNN, Generalized Mean Distance KNN, Mutual KNN, and Ensemble Approach KNN across seven benchmark datasets representing diverse chronic diseases. Performance is assessed using classification accuracy, precision, recall, and F1-score metrics. Experimental results highlight that Adaptive KNN offers significant improvements in accuracy and precision, while Locally Adaptive KNN excels in handling imbalanced datasets. This analysis provides valuable insights for selecting optimal KNN variants, advancing the development of accurate diagnostic tools.