HEKA

Diabetes Detector for Planet Hacks Presenting Project Heka, the health innovation that aims to efficiently diagnose diabetes as well spread awareness. Split up into two components, Project Heka incorporates machine learning in order to detect diabetes with an accuracy of 98%. The machine learning platform determines the best possible algorithm, from Logistic Regression, K-Nearest Neighbors, SVC, Gaussian Naive Bayes, Decision Tree, and Random Forest. The algorithm with the best accuracy is chosen to be incorporated and uses number of pregnancies, glucose levels, blood pressure, skin thickness, insulin levels, BMI, diabetes Pedigree Function, and age in order to make an accurate prediction. After predicting a given input’s diagnosis, the result is then sent to firebase, from which the Heka website displays on the diagnosis page. Firebase acts as the intermediary project database. The website utilizes html, css, and javascript in order to create an efficient user-interface for the patient. All in all, not only does Project Heka spread awareness of a very common health condition, but it also eases the ability to diagnose a patient using technology.