VocaMed

Inspiration

The idea for VocaMed was born in the waiting rooms of Montreal’s overcapacity clinics, watching the growing gap between medical care and medical documentation. We saw firsthand the "Death by a Thousand Clicks," a mounting crisis where doctors and nurses are forced to spend nearly half their shifts tethered to keyboards instead of patients. This isn't just an administrative problem; it's a human one. When a clinician is buried in a screen, they lose the ability to maintain eye contact and build trust with the person in front of them. We were inspired to create an "Invisible Interface" that returns the human element to the ER. We wanted to build a world where the technology acts as a silent partner, handling the heavy lifting of paperwork so that healthcare workers can finally look their patients in the eye again.

What it does

VocaMed functions as a seamless digital relay race across the clinic. It starts at the reception desk with a streamlined intake portal that eliminates the paper trail from the first second. Once registered, the nurse uses a dedicated station to sync vitals directly to a Supabase backend. The moment those vitals are saved, the patient automatically "pings" into the doctor’s digital waiting room. When the doctor enters the exam room, they hit "Start Exam," and our system takes over. By leveraging ElevenLabs WebRTC, VocaMed scribes the physical exam in real-time, transcribing complex medical observations and summarizing them into a clinical report as the doctor speaks.

How we built it

We designed a high-speed, full-stack architecture to handle the pressure of a medical environment. The backend is built with FastAPI, which manages the business logic and secure data flow. We chose Supabase for our database to ensure that data persists across every station in real-time. The frontend was developed using React and TypeScript to provide a safe, responsive interface. The core of the experience, the real-time transcription, was achieved by integrating ElevenLabs’ Conversational AI, allowing us to bridge the gap between spoken word and digital record with almost zero latency.

Challenges we ran into

Our biggest technical hurdle was the "white screen" paradox. In our early builds, the app would crash because our frontend was trying to read patient data before the database had finished its update. This forced us to master TypeScript and implement strict interfaces to ensure the code knew exactly what data to expect and when. We also faced significant challenges with data synchronization; connecting the frontend to the backend required us to move away from simple URL parameters and toward robust JSON payloads to ensure that Supabase stored every vital and note perfectly without overwriting previous data.

Accomplishments that we're proud of

We successfully integrated a multi-member system where Member A (FastAPI) and Member B (React) communicate in real-time to manage patient data. We are particularly proud of overcoming the immense technical difficulty of building a functional AI pipeline that successfully generates structured medical text from raw voice recordings. By orchestrating ElevenLabs and OpenAI we moved past complex formatting errors to ensure clinical speech is transformed into professional, stored documentation

What we learned

As beginners developing through AI prompts, we learned the critical "Same Change" rule, which taught us that any update to the database schema must be manually synced across the backend models and frontend interfaces to prevent the system from breaking. We also gained deep experience in debugging "White Screen" errors. We discovered that using optional chaining (?.) in React and data filtering in FastAPI are essential to prevent NULL or "ghost patient" data from crashing the entire UI. Furthermore, we realized the strictness of data formatting; our encounter with the InvalidConfigurationError taught us to use escapeJSON to handle special characters in AI transcripts so they can be stored correctly.

What's next for VocaMed

Moving forward, our primary goal is to implement User Authentication through Supabase Auth to ensure that sensitive medical data is protected and HIPAA-compliant. We plan to introduce Role-Based Access Control, creating customized dashboards that allow Nurses to focus on intake while Doctors manage the diagnostic summaries. Finally, we aim to expand the database architecture to include more detailed medical fields, such as vitals, prescription tracking, and historical patient trends, to turn this prototype into a comprehensive clinical tool.