Med8d

Inspiration

It’s hard to forget the first time you look at an X-ray or a lab report as a non-technical person. You’re staring at a black-and-white image of a bone or a page full of charts and numbers, and honestly… you have no clue what any of it means. Most of us aren’t doctors. So when we’re handed a radiology report or blood work results, it can feel confusing and even intimidating. You want to understand what’s going on with your own body, but the information just isn’t speaking your language. Drug leaflets add another layer to this problem. They’re often long, dense, and overwhelming, which makes it easy to miss information that actually matters. That feeling of confusion when looking at important medical information and not knowing what to make of it is what inspired Med8d.

What it does

Med8d lets users snap or upload a photo of any medical document (lab reports, radiology reports, prescriptions, or drug leaflets ) and instantly receive a clear, plain-language explanation. Key features: • Snap or upload medical documents for instant explanations • Interactive chatbot that allows users to ask follow-up questions about their document in plain language • Highlights key details users should pay attention to • Supports up to 10 languages for accessibility • Designed to feel friendly and human, not clinical or intimidating

How we built it

Med8d is powered by Google Gemini, integrated through the Generative Google API. One of our main workflow requirements was structured responses, so we configured Gemini to return outputs in a strict JSON format, which made it easy to parse explanations, summaries, and highlighted points consistently. We used Gemini 2.5 Pro, and its strong reasoning and multilingual capabilities made it especially effective for: • Explaining complex medical language accurately • Switching seamlessly between languages • Maintaining clarity without oversimplifying

Challenges we ran into

• Initially receiving unstructured text responses instead of JSON, which broke our processing workflow • Refining prompts to ensure explanations stayed simple, accurate, and non-alarming • Handling diverse document formats and image quality • Balancing medical correctness with plain-language explanations

Accomplishments that we're proud of

• Successfully building a working end-to-end document explanation pipeline • Implementing reliable structured outputs using Gemini’s JSON response configuration • Supporting multiple document types and multiple languages • Creating a product that can help millions of people worldwide

What we learned

• How powerful structured AI outputs are for real-world applications • The importance of prompt design when working with generative models • How multilingual AI can significantly improve accessibility • That impactful healthcare tools don’t have to be overly complex , they just need to be thoughtful

What's next for Med8d

Med8d is just getting started. Our next focus is making the experience feel even more natural and supportive for users. We plan to improve how MedAid understands different medical documents, especially complex radiology reports and lab results, so explanations become even clearer and more personalized. We also want to make the chatbot smarter and more conversational, allowing users to ask follow-up questions over time and keep context across multiple documents. Beyond that, we’re exploring deeper multilingual support to reach more communities. Ultimately, our goal is simple: to make understanding medical information feel effortless , something people can do confidently, without stress or second-guessing.