MediGuide

Inspiration

I just started talking to a girl romantically. She is a doctor by profession and I was thinking to myself, is there something I could do to impress her, something thoughtful. When I stumbled on this hackathon, the idea came into my head: a medical guidance website. I made it for the average Nigerian who lacks the necessary means to communicate their daily medical concerns, for whatever reasons, illiteracy or expensive consultancies. I tried to make the website very accessible to everyone.

What it does

The website has at least four features. Firstly, you have a chatbot where you can talk to an AI doctor and it gives you medical advice like having your own personal consultant. Secondly, if you just want to submit a medical concern quickly, we have a simple form for you to fill out. It also gives you health articles to read and can provide emergency contacts based on your location.

How we built it

I drafted my prompts in plain english, I made sure to capture as much detail as possible for each prompt because of the low amount of credit that we were provided. MeDo was actually able to one-shot the web app based on exactly my intent.

Challenges we ran into

There were no major challenges in particular. I was able to publish with a single click of a button, and my intent for every feature was understood very well by the model.

Accomplishments that we're proud of

I am very proud of the fact that the model was able to implement every feature I requested, especially the geo-location feature that detects a user's location to provide details of available nearby medical services. I also feel that users will find the AI chat doctor very intuitive and helpful in understanding their medical situation.

What we learned

So, I am actually astonished by the capability of MeDo, when I first saw the user interface, I thought to myself that it will be a very weak LLM / Model that will lack severely in reasoning, inference and context compared to market forces such as Claude and Deepseek, but it handled very well. One thing I learned was that you need to treat patient data with a lot of security so, all the conversation is stored on the User PC localstorage and there is an option to delete your conversation once you are done.

What's next for MediGuide

I would definitely go back after this hackathon and build MediGuide out fully to make it more polished, currently it looks very AI-generated. I genuinely enjoyed this hackathon because, even with the limited credits provided, I was able to complete the entire project in one day. I plan to make it a side project, add it to my portfolio, and hopefully find real users who can benefit from it.

Built With

• nextjs
• react
• typescript