D-tect

Inspiration

So, our idea actually started from a personal story — one of our friends spent over $2,000 just for registration and diagnosis, and it turned out to be a common cold.
That got us thinking: in the U.S., it can take weeks to see a doctor, and even simple visits can be really expensive.
We wanted to create something that helps people understand their symptoms quickly and easily, without needing to wait or spend that much money.

What It Does

D-tect is a simple web app where users can enter their basic information — things like age, gender, height, and weight — and describe their symptoms in a text box.
The app then gives back a list of possible illnesses and some preventive or self-care suggestions.
For example, if someone types “fever and sore throat,” it might suggest the flu or strep throat and explain what to watch out for or when to see a doctor.

How We Built It

We built the front end using React and Next.js, and the back end with Node.js connected to Gemini.
Basically, when a user describes their symptoms, our app sends that info to the Gemini API, which uses natural language understanding to analyze it and return clear, human-friendly results.

Challenges We Faced

At one point, our API key wasn’t loading properly in the environment, which completely broke the app.
We spent a while debugging and finally fixed it by adjusting the environment checks and redeploying — definitely a good learning experience.

What We’re Proud Of

We built a fully working prototype in under 15 hours that connects user inputs to an AI-powered, privacy-safe health explainer.
We’re proud that it’s not only functional but also fast, intuitive, and responsible in how it handles sensitive health info.

What We Learned

We learned how to combine LLM reasoning with rule-based logic to make AI more trustworthy and interpretable.
And more importantly, we learned that when building AI for healthcare, clarity, empathy, and safety are just as important as technical accuracy.

What’s Next

Next, we plan to build a confidence-scoring system that checks how reliable each result is by comparing the LLM’s reasoning with our rule-based predictions.
We also want to collect real user feedback to refine the model and improve its accuracy.
Our ultimate goal is to make D-tect a tool that’s smart, safe, and genuinely helpful for anyone who wants to understand their health better.