Medical Chatbot Unimedic 🤖💬

Idea: The idea that motivated the development of the project was to facilitate medical analysis for patients and doctors, avoiding unnecessary medical appointments.

PytorchFastAPIReact Native

Live Demo Source Medtorch

Motivation

This project was based on the medtorch repository (winner of the 2020 Pytorch hackathon). Both the application and the deployment were very interesting processes, showing us in a challenging way how we could tackle real-world problems with the help of various Deep Learning tools.

Features

⚛️ Interface created with React-Native.
🍃 Authentication using Mongo DB.
🌀 Chatbot powered by GiftedChat.
🔥 Powered backend with models built with PyTorch.

Screens

Intro

Welcome uniMedic Let's get started!

Patient interface (chat)

Invalid Image Result with segmentation

Doctor interface

Invalid Image Result with segmentation

Invalid image Result with segmentation

Diagnosis PDF

Invalid image

Installation & Setup

Follow these steps to set up your React Native environment. I also recommend that you follow this tutorial in order to properly configure Android Studio. To successfully connect your mobile phone and/or emulator, you need to install ADB and set your environment variables.

Once you have installed the packages with npm i, you can run the following command:

npm start

This will automatically open Expo in your browser. I recommend you use Expo Go to run the app directly on your phone without using the USB cable.

Next steps

🐼 More PyTorch models for vision tasks (lungs, breasts).