Diabetest: Mobile Application for T2DM Detection Among Filipino Young Adults Using Facial Texture

Diabetes is increasing in both prevalence and incidence worldwide. The International Diabetes Federation (IDF) showed that in 2021, 15.51 percent of the total adult population have diabetes. Current invasive and costly DM detection methods raise the need for more available and accessible non-invasive detection systems for diabetes given the prevalence of diabetes in Filipino adults. The research focused on the development of a mobile application employing Facial Texture Analysis as a novel method for detection of diabetes among Filipino adults, ages 21–40. The mobile application, which was built using Python, allows the user to upload a facial image which the application processes. The photo will undergo resizing, facial block extraction, texture feature extraction using the Gabor filter, and diabetes classification using the Support Vector Machine model, trained with a 70:30 training-to-testing ratio, which was stored in a web server.