Overview

This is the code for this video on Youtube by Siraj Raval on how to build a biomedical startup. I basically saw these Colab demos of this BioBERT model answering medical questions pretty acurately and got hype. I thought, why not apply this type of interface to that as a backend, paywall it, and hey! You've got yourself a business, serving the underserved st scale at a very low cost. This app works, but there are still TODOs, i'll list them below.

Usage

Install flutter. Then, install all dependencies and run using these commands. I prefer running it using Android Studio, but terminal works too.

flutter packages get
flutter run

:tv: Video demo

TODOs - Make a PR if you fix any

• Take the model here and integrate it into the app. My previous healthcare app nicely integrated tensorflow lite.
• Parse the input using keyword recognition. You can use an existing API like this to categorize words as symptoms, medications, or side effects. Train a new model on a health dataset for the most accuracy, have it learn to classify those 3 keywords from a given corpus. Once the input is parsed properly into those categories, it can be fed into the existing model.
• Finish Integrating Firebase
• Finish Integrating Payments

Credits

rxlabs and the Doc Product team. Also the Flutter team.