Abstract
With diabetes mellitus being one of the most common diseases in the world, comes with various complications if one were not to get it treated in time. Hence, it is important for one to have quick access and easy to use diagnostic equipment with high accuracy for early detection. This thesis will be exploring non-invasive methods to predict a patient’s risk of pre-diabetes and diabetes and exploring it’s impact and usability through a prototype. Using Acanthosis Nigricans and slow healing cuts and sores as symptoms to detect, proven Deep Learning models to perform classification on these symptoms. Furthermore, this thesis proposed a full architecture prototype, covering from hardware designs to software architecture. The prototype can potentially be a product which serves the global population in the early detection of prediabetes and diabetes, reducing the possibility of patients getting complications due to the disease