Comparative Analysis of Machine Learning Models in Early-Stage Diabetes Prediction

As the implementation of machine learning (ML) techniques in the medical industry is growing, utilizing ML models for prediction is crucial for diabetes prevention and treatment. Employing data extracted from the early-stage diabetes risk prediction dataset on the Kaggle platform, consisting of 520 samples, this study established the confusion matrix, applied four ML techniques: logistic regression, decision trees, random forests, and k-nearest neighbors, and compared the application outcomes of several models to assess the effect. According to the study, the four models performed well in predicting diabetes in its early stages. Decision trees and random forests achieved 99% and 98% accuracy, respectively. The interpretation of results provides favorable evidence supporting the application of ML in the early diabetes diagnosis. It manifests that ML has great potential in the early forecasts of diabetes. The research, however, still has the problem of a tiny sample size and insufficient model training. Increasing the incorporation of real-time data in order to strengthen the model's flexibility and long-term prediction capabilities is an excellent way to tackle the matter.