Abstract:
"
The DiabiPred system is designed and implemented to identify the risk of diabetes in the
patients who use Statins medications to lower the blood cholesterol levels. The researches
prove that there is a connection between the using of Statins and the risk of the diabetes
for some unknown reason, but the risk is important that the FDA has issued warnings on
the Statin labels about the Diabetes and the risk associated with it. Therefore, this system
has taken that issue as the point of consideration and the model and the project is designed
to get informed about the risk of getting diabetes in the patients who use the Statin
medications. The statin usage cannot predict the diabetes risk all alone, therefore attributes
such as age, weight, height, working status, exercise routine, diet and sex are taken as the
attributes irrespective of the statin name, the dosage and the period of taking statins. The
machine learning tools are used in the project along with the image processing. The
process of the prediction is to input the statin included prescription to the system and the
other attributes can be filled by the user. A dataset of 1844 responses is taken in order to
train the model for the prediction. The prediction accuracy system is up to 79.4% since
the approach was taken with the Random Forest classifier, after testing the model with
SVM, KNN, Extra Trees and Bagging Classifier. The application is designed as a mobile
application where the data frontend of the mobile application was designed by flutter and
dart."
