Inspiration
When I started this project I was confused. I did not understand the patient at first. I judged it on instinct and said I would build a tool.
But as I did more research, and as I saw people suffering from this disease, my heart felt so tight. It made me see how wealthy we are, and how we never appreciate the things we have until they are gone.
It made me more considerate of people with ALS, who wish for one more moment with their kids, one more moment in the kitchen or at school, while the rest of us doomscroll on our phones and think about how miserable our lives are.
There is no cure for ALS. So the only thing anyone can actually give Alex is time, and time is exactly what he was not given. He cannot get back the two years he lost. I did not want that to happen to anyone else.
What it does
A two-minute screen for a weak hand or a dropping foot that has been blamed on the wrong thing.
On 9 December 2024 Alex's hand surgeon wrote three things in one paragraph: a normal nerve test, a shrunken muscle at the base of the thumb, and a shrunken muscle between thumb and index finger. Those last two run on two different nerves. Carpal tunnel squashes one of them, so it can shrink one, never both. It was written down. He was operated on six weeks later and diagnosed 19 months after that.
A year later the same error happened to his leg: a clean back scan, a calf 2 cm thinner than the other side, and six months of physiotherapy for an ankle problem he did not have.
The tool takes what a clinician found and returns one of four verdicts, up to "do not treat this as a trapped nerve, send to a neurologist urgently."
There is also a second mode for patients and families. Plain questions you can answer by looking at your own hands, no tests needed, and it writes out the words to take to a doctor. That matters most for people who cannot afford a nerve test or do not have a neurologist nearby.
How we built it
One self-contained HTML file. No framework, no backend, no dependencies. The rule engine is about sixty lines. It runs on any phone and works offline.
The rules came from reading the case file closely rather than from a dataset, because nothing was missing from Alex's notes. Every test was done. The problem was that three facts never got put side by side.
Challenges we ran into
Not understanding the patient at the start. I spent a long time on a different idea before I understood what the case was really about, and I had to drop it and start again on Saturday night.
Working solo, which meant scoping it down to one thing I could actually finish.
And the false alarm problem. Carpal tunnel surgery is very common and ALS is rare, so if the tool fires too easily a neurology clinic drowns in referrals. The stop signs are deliberately written to be rare. If a records review showed they are common, the tool should not be used.
Accomplishments that we're proud of
It replays all five of Alex's real appointments and fires at every single one, including the first.
It covers both of his misdiagnoses, the hand and the foot, not just the obvious one.
The patient version. Someone who cannot reach a neurologist can still find out that their story does not fit what they were told, and walk in with the right words.
And I am proud that I said out loud what would make my own tool wrong. The page has a section on how it should be tested and what result would mean it should not be used.
What we learned
I had heard of ALS before, but I did not know it could be like this, or how many people in Canada live with it.
I learned that the delay was not caused by a missing test. Everything was done and everything was written down. It was a thinking failure, not a technology failure, and you do not fix that with more technology.
I also learned that ALS Canada already has a referral tool, and that the Gold Coast criteria already exist for confirming the diagnosis. Both of them only start working after someone has thought of ALS. Alex spent 20 months before anyone did.
What's next for Wrong nerve
I want to keep going after the hackathon. I will finish the idea and one day turn it into a proper app.
Before it could be used for real it needs testing: ten years of carpal tunnel referrals matched against ten years of ALS diagnoses, with the screen run on the old notes. Four numbers — how many it catches, how many false alarms, months gained, operations avoided.
It is true that things like this already exist. The point is to make it better. Just as we learned what happened to Alex, we make sure it does not happen again.
When I joined this hackathon my goal was to find a team and win a prize. By the end my goal was to find a solution, and to keep working on it after the hackathon is over.
Built With
- claude
