Diabetes Diagonistic Prediction using MachineLearning

Abstract

It is very important to detect diabetes early to prevent the occurrenceof health issues in the long run, yet many individuals in developing nationsdo not have the opportunity to be checked in the proper way. Inthispaper, a machine-learning approach to predicting diabetes basedonmedical and lifestyle data that is relatively simple to detect is displayed. Our sample was a collection of 1,500 patient records, which consistedof diabetic and non-diabetic patients. Once the data had been cleaned, andthe features established, we trained some machine-learning models andtested them with each other. The best model achieved 97%as theaccuracy rate and it is indeed high despite the lack of enormous data. These results demonstrate that machine-learning may provide a up-to- date and trustworthy instrument to spot early risks in diabetes andit does not require the advanced complexities of deep-learning systems. The paper highlights the presence of inexpensive and data-enhanceddiagnostics that may assist in accessing medical assistance to individualssooner and strengthen healthcare judgments in resource-scarce regions.