ADVOKAT | Devpost

Inspiration

Every year, hundreds of thousands of lives are lost because patients aren't diagnosed in time. Not because doctors don’t care, but because symptoms were missed, misunderstood, or never clearly communicated.

Women with endometriosis wait up to 10 years for a diagnosis.¹ A four-week delay in cancer treatment can increase the risk of death by 10%.² And over 70% of ovarian cancers are diagnosed late, often because early signs like bloating or pain go unrecognized.³

We realized this isn’t just a medical problem, it’s a communication problem. Many patients struggle to articulate their experiences in a way that doctors can act on. That’s why we built Advokat.

What it does

Advokat is an AI-based web app that helps patients organize their symptoms and health history into a clear, medically relevant report. Users answer a series of guided questions about their symptoms, medications, medical history, and concerns. Advokat then generates a structured summary of symptoms using AI.

Here’s what it provides:

-Key symptoms and health details -Important points to raise during the appointment -Suggested questions to ask the doctor -Recommended tests to consider

However, if patients want a more detailed report, they can answer additional follow-up questions. We generate questions using AI specifically to fill in gaps in the patient reports . Therefore, they can receive the most detailed report!

Patients can also edit their inputs and download the full report, ready to print or bring to their next visit.

How we built it

-Front End: React -Backend: Django & Auth0 for authentication -AI Model: Lunon’s API

-Data Sources: Powered with the help of KNOT

We built Advokat using Next.js and TailwindCSS for the interface and layout. To handle AI-generated summaries, we integrated OpenAI through a custom API route. We designed and implemented features like:

-The guided symptom questionnaire -Editable follow-up questions -Exporting reports as .txt files

Challenges we ran into

One of the biggest challenges was formatting the AI-generated summary to be both readable and medically relevant. Another challenge was designing the dynamic question flow: figuring out how to ask just enough without overwhelming the user while updating the summary in real-time as responses changed. Balancing technical functionality with user clarity was tough, but it pushed us to simplify the experience without losing depth.

Accomplishments that we're proud of

We are proud that we were able to create a fully functional medical tool in less than 36 hours. All of our team members entered this hackathon with the goal of building software that could make a real impact, and we believe we’ve done just that. Advokat helps all patients (ourselves included) communicate more clearly with their doctors and advocate for their health more effectively.

Advokat is designed to support earlier diagnosis, reduce the chances of being misheard or overlooked, and help patients walk into medical visits with clarity and confidence.

What we learned

-Great research is essential for building a great project -Technology tools are only as powerful as how they’re used

-We deepened our expertise in: JavaScript and React TailwindCSS Working with AI APIs Building with accessibility and clarity in mind Implementing Auth0 authentication systems

What's next for ADVOKAT

First, we want to get Advokat into the hands of patients! Our goal is to solidify our systems and publish this project on the web.

In terms of features, we plan to add more detailed medication tracking to make Advokat a part of daily life, because true understanding comes from consistent health monitoring.

We also aim to integrate with existing medical systems, enabling patients to upload medical records and automatically send alerts to doctors. The sky’s the limit! Future enhancements include:

-Lab result input -Voice-based symptom entry -Templates for chronic conditions

Additionally, we hope to partner with clinics and advocacy groups to reach patients who struggle to navigate the healthcare system.

Citations