HelpMyHealth

People often Google symptoms but get unreliable answers. My AI app interprets symptoms using trusted medical sources and tells you: How to treat it yourself with self-care, a doctor, or emergency.

Comment

Inspiration

Most people Google their symptoms when they feel sick, but search results are often confusing, unreliable, or unnecessarily alarming. It’s easy to panic over harmless symptoms or ignore serious ones. I wanted to build something that helps people make a decision, not a diagnosis — combining the clarity of first-aid guidance with the transparency of medical sources.

What it does

Help My Health is a health triage assistant that interprets what a user describes and determines the appropriate level of care: self-care, doctor, urgent care, or emergency. Instead of generating advice, it gathers guidance from trusted medical sources and explains what actions to take and why.

How I built it

The project uses a hybrid approach: Frontend: chat-style interface for describing symptoms Backend: FastAPI service Medical guidance retrieval: NHS and MedlinePlus pages Research evidence: PubMed studies ranked by reliability AI component: Gemini used to analyze sources and improve reliability evaluation Emergency detection: identifies red-flag symptoms and provides location-based emergency numbers

The goal was to let AI interpret and organize information while keeping medical advice grounded in trusted references.

Challenges I ran into

One of the biggest challenges was safety. Medical apps can’t behave like typical chatbots; generating advice freely is risky. I had to design the system so that the AI helps understand and evaluate information, but doesn’t invent treatment instructions.

Another challenge was extracting useful steps from real medical websites. Pages are written for humans, not machines, so I had to handle inconsistent layouts, missing sections, and unreliable formatting.

Integrating AI also required careful prompting to ensure it judged reliability instead of giving medical advice.

Accomplishments that I'm proud of

What I am most proud of is how I was able to get my idea of what I wanted in a general sense, but there is always more to add and more to improve, and still, I am generally happy with my project.

What I learned

What I learned from creating this project is how much fine-tuning goes into finalizing and making the project work, not just by using AI to help with creation.

What's next for HelpMyHealth

What's next for HelpMyHealth is just probably going to stay as it is for the meantime, since it is done on such a small scale with only a dozen websites used to take information from. Maybe in the future, if I see potential, or if anyone sees potential with a similar idea as this, they can create a fast help webpage or app and incorporate all health data and give the best recommendation from an AI stance, without hallucinating.

Share this project:

Updates