d-bhatta/diabetes-prediction-android

An android app with a python backend to do diabetes prediction

Diabetes-Prediction-Android

An android app with a python backend to do diabetes prediction

Idea

The android app uses a python backend to do the prediction based on an aggregation of models. The results and data are written to a database. The app connects to a server through a URL. A base request handler handles the request here, through a socket. Then, it calls a Classification function. The results of the classification function are sent to the android app via callback function. Then, the results are written to the database with a user identifier.

Classification function

The function returns the aggregate result of several models, whose results are summarized as text and returned.

Callback function

The callback function returns to the app first a response of 200. If there is no result then a 400 is returned. This is followed by a string containing the results.

400

The app displays an error and handles appropriately.

200

The app receives the result and displays it.

The app

There is a splash screen at first. Then the app has a login/registration id page, along with server address. This is used to login. The app then stores the data from the user. It sends the data to the server. It displays a result.

Improvements

use flask instead of the base request handler: Search for 'simple flask app' in google.
Attach a weight tracker to the app.
Display list of improvements in health the user can make.

List of To-dos

• Train the models and save them

  1. Train the models and save them as files.
  2. Load and test them for predictions.
  3. Create a class that classifies input data.
  4. Summarize the aggregate result of several models as text in the class and analyse accuracy.
  5. Return the accuracy and the predictions.
  6. Test it with stubs.

• Create the app with stubs

  1. Create a nav drawer page as the main page.
  2. Create the splash screen and test it with stubs.
  3. Create a login and registration page and test it with stubs.
  4. Create a form page for user data input
  5. Create a result page and test it with stubs.

• Create the webs server with stubs

  1. Create a flask app. If that doesn't work, create a simple HTTP base handler server.
  2. Set it to handle some requests.
  3. Test the requests with stubs.
  4. Set it to return responses.
  5. Test the responses with stubs.
  6. Create the Classification function.
  7. Attach it to the Classification function.
  8. Test requests and responses with stubs.

• Attach the server to the app

  1. Attach the server to the app.
  2. Test for sample input data.
  3. Remove bugs.
  4. Publish as Alpha version.

• Test the app

  1. Test it for a wide range of input data.
  2. Find its limitations.
  3. Document it's expected input and output.

• Create the documentation

  1. Create the project report.
     1. Use README.md as a reference to create the report introduction and methodology.
     2. Display statistics and figures about the data.
     3. Show mockup figures.
     4. Display expected results for sample inputs.
     5. Talk about future imporvements.
     6. Conclusion
  2. Create the project PowerPoint.
     1. Turn the report into a PPT.
     2. Divide roles of speaking.
  3. Create a paper on the software.