Issue Details

HARNESSING MACHINE LEARNING FOR EARLY PREDICTION OF DIABETES ONSET IN AT-RISK POPULATIONS

ABSTRACT

The purpose of this research is to examine how machine learning algorithms might aid doctors in diabetes risk assessment and early diagnosis. The investigation made use of a dataset collected from Ninh Binh people who were Vietnamese and had a history of type 2 diabetes. The best classification technique for the dataset was determined using a variety of techniques, including K Neighbors, Ada Lift, Calculated Relapse, SVC, Random Forest, and Choice Tree Classifier. Results indicated that the Random Forest Classifier computation had the highest potential, with a precision rate of 100% and a cross-validation score of 0.998. Applying the chosen model on a new dataset that removed 67 people from the original allowed for a more thorough evaluation of its effectiveness. The algorithm had a 94% success rate on this dataset. Class 1 (diabetic) probability show that it did a great job of predicting whether individuals will develop diabetes. This innovative approach demonstrates how machine learning algorithms may help clinicians with patient care and diagnosis by providing a systematic and measurable technique to detect diabetes early on and evaluate risk. Giving patients access to their diabetes score and probability estimates may help them better understand their condition. This information is urgent in forestalling or easing back the movement of illness since it engages patients to go with educated decisions and urges them to embrace solid way of life ways of behaving. This exploration shows how significant machine learning is for medical services to improve patient consideration and wellbeing results over the long haul.

FULL TEXT