Medisense

Inspiration

As believers that artificial intelligence is a great tool all around, we wanted to create a tool for medical professionals. We wanted them to be able to corroborate diagnosis using artificial intelligence to be more confident.

What it does

It uses a random force classifier model to calculate the risk factor for a patient to have a certain disease. A doctor uploads a patient's records and selects what disease they are diagnosing them for. The model determines if the patient is at high risk for the specified disease. If a doctor diagnoses a patient for a disease and sees that the patient is high risk for that disease, they can be more confident in their diagnoses. If a doctor diagnoses a patient for a disease and sees that the patient is not high risk for that disease, they may rethink their diagnosis.

How we built it

We used NextJS as a fullstack framework to handle the frontend and the backend of the application. We used sklearn to train the model using a data set from kaggle and flask to send the data to the NextJS application from the python code.

Challenges we ran into

We had a hard time deciding which model was the most efficient and parsing the data. Another issue was passing a file from the frontend to the backend. In the end we worked through it and were able to choose an efficient model with parsed data and functioning file sending.

Accomplishments that we're proud of

We're proud of how we were able to work across different frameworks and languages to make a synchronous apps.

What we learned

Each person learned different things. Our backend engineers learned some frontend development, our frontend devs learned backend, and our fullstack devs learned a bit of both!

What's next for Medisense

Adding doctor signups to keep track of all their patients risk of disease over time, creating more detailed graph outputs, and increasing the data set the model was trained on.