Class Disproportion using a Variety of Machine Learning Approaches to Predict Clinical Risk Factors for Diabetes

Diabetes is the most prevalent and quickly spreading disease that shortens life expectancy. It affects a great number of people annually from all age groups. posing a strong impact rate, it makes the first diagnosis more significant. Diabetic carries additional difficult consequences such as cardiovascular disease, renal failure, stroke, harm to important organs, etc. Earlier Diabetes diagnosis lowers the risk that it may progress into a severe and ongoing condition. Determining the frequency of diabetes in medical analysis is facilitated by the identification and assessment of risk factors for certain spinal features. Future consequences are less likely when diabetes is detected early on and its prevalence is measured. This study used the combined NHANES dataset from 1999–2000 to 2015–2016. Its goals were to use multiple supervised machine learning algorithms to identify abnormalities and to analyse and determine possible risk variables connected to diabetes by means of ANOVA and logistic regression. The experimental results, which corrected for class imbalance and outlier issues, indicate that bloodrelated diabetes, age, cholesterol, and BMI are the main risk factors for developing diabetes. Along with this, the random forest classification approach yielded the greatest accuracy score of 90.