Through this pandemic, we noticed the length and the inevitability of this disease. As we wanted to look into the “future”, we created this application that allows better communication between patients and doctors.

What it does

Our app allows doctors to sign in and upload x-ray scans, and our machine learning model will diagnose whether the scan reflects signs of COVID-19, Tuberculosis, or Pneumonia. Our model also informs the doctor if the x-ray scan shows no sign of infection. Our app allows doctors to also keep track of all uploaded scans and to share diagnoses with their patients.

How we built it

We built this using React Native, which is a framework that uses JavaScript that enabled us to perform all the front-end engineering of our application. Then, we managed all of our data into collections using Firebase, such as storing user data and all uploaded scans into Firebase's Cloud Firestore Database. Firebase also allowed us to implement user authentication. Next, we used Tensorflow.js to implement our machine learning model and deploy it onto our React Native app.

Challenges we ran into

One particular challenge we ran into were implementing user authentication. User authentication would work for one of us and not the other, so often times only one person could work beyond a certain point. This was especially challenging for it turned a team project into almost a solo project, which made it twice as difficult to properly execute our app.

Accomplishments that we're proud of

We're especially proud of our machine learning model. We were surprised that we could quickly load accurate results and securely store them into our database. Also, we're proud of even finishing this app, as user authentication would only work on one of our phones, so it almost became a solo project at one point. This made it very difficult to complete the project and it took much longer than we hoped.

What we learned

We learned how to collaborate together even when we faced severe problems that impeded one of us from working on the app. When user authentication stopped working specifically for one of our phones, this almost became a one-man mission, but we learned to persevere through all the challenges and worked even harder to get finish the app. The main takeaway from the development process for us was how to work together even if we literally cannot.

What's next for ChestRay

We hope to make our machine learning model more applicable by adding more classes (more respiratory infections) when training our model. This broadens the scope of our model, as we were only able to use our model for 4 different classes, which were COVID-19, Pneumonia, Tuberculosis, and healthy lungs. We next want to incorporate other infections such as Bronchitis, Laryngitis, and the Flu.

Built With

Share this project: