MediVyne

Inspiration

In diagnosing a patient with an illness, medical professionals go through the lengthy process of recording symptoms, administering tests, and narrowing down the possibilities. We aimed to create an app that would serve as a physician's partner in investigating a patient's symptoms and coming to the right diagnosis. This tool could assist medical professionals in tracking symptoms and test results visually, and better serve their patients by providing a visually impactful and interactive tool for the differential diagnosis process.

What it does

The app allows doctors to textually input patient symptoms, and creates a graph based on those symptoms. To the left of the root--clickable nodes suggest possible diagnoses, generated by OpenAi's GPT-4o mini, and serve medical research papers relevant to the doctor's input. To the right of the root--nodes suggest relevant medical testing methods (Complete Blood Count, PCR, skin biopsy), and by clicking one of these nodes, the medical professional can record the results of that test. This will then create new left and right "vines", with nodes again suggesting diagnoses, tailored by the new information received.

How we built it

Our team divided ourselves into a frontend and backend team, and research tools, and designed the UI and backend architecture of our app first. After settling on the tools we would use and setting up a repo, we began building!

Challenges we ran into

Merge conflicts
Scope creep ## Accomplishments that we're proud of
Building a functional web app with a visually interesting and dynamic interface.
Learning new technologies ## What we learned
New technologies: Flask, Next.js, OpenAI integration
How to plan, implement, and deploy an app within a day-and-a-half
Working in a team can be messy but by level-setting expectations, learning about team skills and honing in on them, and individually striving for learning at every step, you have great chance at achieving what you set out to do. ## What's next for MediVyne Each of us hope to improve the app in our own unique ways, using our work from this hackathon as a prototype for something grander. Critically we hope to improve recommendation validity by adding the use of medical terminology libraries and ML fine-tuned for medical applications.