Design and Analysis of an Early Diabetes Mellitus Detection Using OpenCV

This technical report summarizes the whole developing process of this project. The project is about developing a web application to detect early diabetes mellitus. The problem statement for this project is the lack of convenience in detecting diabetes. The only way to diagnose diabetes is through blood test but blood test is not recommended to conduct frequently. There is also lack of application to help to detect diabetes. To solve the problem, we have found the most accurate machine learning algorithm which is Random Forest to detect diabetes. Random Forest algorithm will be the prediction model and will be deployed into the web application. The web application is developed by using Visual Studio Code with the assistance of Streamlit framework. To develop this web application, there will be two parts, which are prediction model development and user interface development. Prediction model development involves choosing the most suitable machine learning algorithm to be the prediction model. User interface development consists of the UI of the web application. For 30-year-old male without symptoms, the results are 1 for negative 0 for positive. With only polyuria, the results are 0.95 for positive and 0.05 for negative. With only polydipsia, the results are 0.87 for positive and 0.13 for negative.