Inspiration

I kept thinking about how epilepsy patients spend years on the wrong drug. Doctor picks one, it doesn't work, you try another. But the data to get it right the first time already exists — your genetics, your blood work, your seizure type. Nobody's using it at the moment of prescription. That's what NeurX fixe

What it does

You input your seizure type, blood work values, and genetic variants (CYP2C9, CYP2C19, HLA-B*15:02). NeurX runs a clinical scoring engine across 7 anti-epileptic drugs and gives you the best match — with the starting dose, target dose, frequency, and therapeutic window. Alternatives are tappable cards with full dosing. Clinical details are hidden behind collapsibles so patients see plain English and doctors can verify the reasoning.

How we built it

Vanilla HTML/CSS/JS, one file, no frameworks. The core is a hand-built scoring engine — each drug gets scored based on seizure type match, then penalized for CYP interactions, HLA risk, liver stress, kidney function, and sodium levels. All ranges are validated against real clinical data.

Challenges we ran into

Building clinical logic without being a doctor. I had to actually read CPIC guidelines and PharmGKB annotations to make the scoring defensible. Also figuring out how to design for two audiences — patients and doctors — at the same time.

Accomplishments that we're proud of

What I learned

One gene variant can completely change how your body processes a drug. Most patients never get tested for this before their first prescription. That gap felt worth building around.

What's next for NuerX

Expand to other conditions, add lab report PDF upload, build a doctor dashboard, integrate with real pharmacogenomics services like GeneSight.

Built With

Share this project:

Updates