Link : https://diabetespredictor1.azurewebsites.net/
This repository consists of files required to deploy a Web App created with Flask on Microsoft Azure.
The project helps the user to identify whether someone is suffering from diabetes by simply inputting certain values like BMI, Glucose level, Blood pressure etc. with the help of a Kaggle database.
By using the statistical data about how certain aspects like BMI, Glucose level, Insulin level, age etc. impact if an individual is prone to diabetes or not, the project will be able to tell the user if the person has diabetes or not by entering those values. So in a way the project will help in monitoring the likelihood of someone developing diabetes. The project can be extended to include other diseases prediction as well.
Health
Diabetes Predictor
Medical diagnosis is a very essential and critical aspect for healthcare professionals. In particular, classification of diabetics is very complex. An early identification of diabetes is much important in controlling diabetes. A patient has to go through several tests and later it is very difficult for the professionals to keep track of multiple factors at the time of diagnosis process which can lead to inaccurate results which makes the detection very challenging. Due to most advance technologies especially machine learning algorithms are very beneficial for the fast and accurate prediction of the disease in the healthcare industries.
The project has three primary stages: creating the model, training the model, and testing the model. Creating the model entails further sub-activities: obtaining, analyzing, and preparing data. Train the model involves selecting and implementing a two-class averaged perceptron. Finally, predict the patient's diabetes by testing the model.
The project was created on Microsoft Machine Learning Studio (classic) using the Azure Open Dataset - Pima Indians Diabetes Binary Classification dataset. Further, the web service is set up and deployed to publish to Gallery. The accuracy achieved is 78.6%.
This approach can help physicians forecast individuals at risk of developing diabetes in the future and provide them with essential preventative measures.
The dataset is originally collected and circulated by “National Institute of Diabetes and Digestive and Kidney Diseases” which is available at Kaggle in the name of Pima Indians Diabetes Database. The main objective is to predict whether a patient has diabetics or not, based on the diagnostic measurements gathered in the database.