Detector

What it does

There are type of diseases which can be managed with combination of lifestyle changes, medicine & in very rare cases surgery.With right prediction this disease type can be reduced by taking proper action and the functioning of the body can be improved, this will help to prevent any high cost surgical treatment and other expenses.

In this project I had implemented two machine learning models which will predict the chances of the heart attack and diabetes on the parameters of the patients given as a input.

In addition we had implemented it into node-red for the UI with a password less SAWO auth to make it easy for everyone to use it.

How we built it

In this project, our machine learning model predicted class 0 or 1, based on the 8 parameters given in the dataset. Here we have used 3 folds cross validation method to split training and test data. Our data set split in to Training data - 90% and Test data - 10%. Our machine learning model has chosen Binary Classification as Prediction type and Accuracy as Optimized metric based on the output parameter 'class'. Once the Experiment Results completed we saved two models of XGBClassifierEstimator with less accuracy and high accuracy. After that we deployed our model using the 'online' deployment type and test our predicted data. In addition, we started Node-Red service and imported the Json file as new flow. We have installed Node-Red dashboard for unavailable nodes. Here we have designed the form for User Interface and changed the API key and variables in the pre-token node. As well as, we have given Endpoint in the http request node. Then, we deployed our flow, checked the debug messages. Finally, we have tested our data in User Interface and predicted the class successfully.