Predicting diabetes with multivariate analysis an innovative KNN-based classifier approach

Diabetes seems to be a severe protracted disease or combination of biochemical disorders. A person's blood glucose (BG) levels remain elevated for an extended period because tissues lack and non-reaction to hormones. Such conditions are also causing longer-term obstacles or serious health issues. The medical field handles a large amount of very delicate data that must be handled properly. K-Nearest Neighbourhood (KNN) seems to be a common and straightforward ML method for creating illness threat prognosis models based on pertinent clinical information. We provide an adaptable neuro-fuzzy inference K-Nearest Neighbourhood (AF-KNN) learning-dependent forecasting system relying on patients' behavioural traits in several aspects to obtain our aim. That method identifies the best proportion of neighborhoods having a reduced inaccuracy risk to improve the predicting performance of the final system. •The work introduces an adaptable fuzzified K-Nearest Neighborhood (AF-KNN) classifier for advanced diabetes prediction.•Fuzzy rules are dynamically altered based on the extracted characteristic features enhances classification and prediction.•It determines the relationship between blood pressure and insulin of the biochemical aspects of diabetes metabolism.•The high level of accuracy (99.32%) consistent across two distinct databases, TLGS and PIDD.