Prognosis the Risk of Early-Stage Diabetes Using Machine Learning Techniques

Diabetes is a prevalent health issue on a global scale, with an exceptionally high incidence rate observed in Bangladesh. The condition is characterized by persistent hyperglycemia, which is having high blood glucose levels in an individual. In addition, it is known to be a contributing factor to various health issues such as visual impairment, renal dysfunction, myocardial infarction, and cerebrovascular accident. The main aim of this study is to evaluate the prognostic value of early prediction of diabetes disease by examining the symptoms exhibited by diabetes patients in Bangladesh. Early prediction can save both money and a patient's life, which is our motive. We have collected 5800 observations with 17 attributes from diabetes-suspected individuals, where 5118 pertain to veritable cases, and 682 are diabetes-negative instances. Several data preprocessing techniques were applied to our dataset to prepare data for machine learning algorithms. We have applied five machine learning algorithms with four performance measurement metrics to assess the performance of those algorithms. Amongst the five distinct machine learning algorithms, the Random Forest algorithm exhibits the highest level of accuracy, reaching 98.22%. Therefore, it can be inferred that the Random forest-based classifier outperforms its counterparts.