HAMM | Devpost

Homepage
Dashboard
Example meeting page

Inspiration 🩺💻

Healthcare disparities have been a prolonged issue such that underserved/minority populations are not able to get equal healthcare access. Some social determinants that prevent patients from getting quality treatments are language barriers, economic status, insurance status, education levels, transportation, etc. Based on these factors, we think language barriers are one of the leading factors in health disparities. By implementing various tech stacks and an AI model we envision our project to be able to annotate language in real-time, and schedule reoccurring notifications or text alerts for ongoing medication treatments or follow-up appointments (for lab results and regular clinic visits) due to varying circumstances. In return, this enables doctors to treat their patients accordingly without a need to converse in another language and increase efficiency with healthcare management.

What is 🍖HAMM🍖?

HAMM is a website that acts as a medical scribe and translator to minimize the language barriers and second brain for doctors and patients. By having a scribe tool that can take notes automatically after the input of audio, we implement text alerts functions for those that “follow-up appointments” or “medication refills” are mentioned during the meetings. HAMM works as:

A Medical Scribe: take notes on the meeting.
A Translator: translate notes into different languages that fit the preferences of the patients.
An EMR (electrical medical record): stores the generated notes for admins and patients to review afterwards.
A Case Manager: Keep track and send text alerts to patients that need follow-up appointments and medication refills.

How we built it 👨🏻‍💻 👩🏻‍💻

Technologies that was used:

Frontend: TypeScript, Next.js, React, Tailwind CSS, shadcn/ui, and Drizzle ORM.
Backend: Python, FastAPI, Deepgram, Twilio, OpenAI, and Upstash QStash.

We also used SQLite (Turso) and PropelAuth on both the frontend and the backend.

Challenges we ran into ❌

The major challenge that we ran into was the scale of the project, we needed to narrow down the scale because of the time limits of 24 hours and funding. There were functionalities that we wanted to do but were limited because of the time and funding issues, such as Zoom API and Twilio two-way communications.

Accomplishments that we're proud of 👍

Implemented and fine-tuned a model, benefiting underserved communities and advocating for populations that experience healthcare disparities.
Bridged the gap between AI and healthcare management systems, making huge improvements in healthcare management workflow efficiency.

What we learned 📚

One technical thing that we learned throughout this hackathon is the Zoom API. Starting off, we wanted to use the Zoom API to create plugins on the Zoom App Marketplace so that we could further manipulate the audio file to our objective. While we managed to login to Zoom from our hackathon website, we were unable to further get the recordings from Zoom as it requires a payment plan. Through this technical experience, we learned how to configure the Zoom API for login purposes.

Another thing that we learned throughout this hackathon is how to acknowledge everyone's strengths and weaknesses. Throughout this hackathon, we've gained insight into the importance of recognizing both strengths and weaknesses within our team. Comprised of individuals from diverse disciplines, we've discovered the connections between different fields and the vital role communication plays in driving substantial societal advancements.

What's next for 🍖HAMM🍖?

The next challenge that we would like to accomplish is to generate translations and notes during live meetings via zoom API, and be able to let patients connect to our medical AI that can answer their general questions.