CareNow

You break your hand. You need care now, but you’re in shock and don’t know how to get treated as quickly as possible. CareNow sorts this out in a way that’s simple and accessible.

Comment

CareNow: Choose the Hospital that Will Treat You First

Inspiration and Solution:

Waiting rooms suck. Being in pain also sucks. We’ve all searched for the nearest hospital on google maps before, but that’s not the only factor. Waiting times have gone through the roof in recent years and subsequently can vary by a matter of hours from place to place. So how do you choose the hospital?

We wanted a one‑click tool that gives a clear, actionable answer: where should I go to get seen fastest? The problem: find the most accessible emergency care. Our solution, Care Now, ranks nearby A&Es by total time to treatment: real travel time plus predicted wait time. It’s one click, no login, no settings.

How It Works:

  • Finds nearby A&Es using NHS API
  • Computes commute time using Google Maps API
  • Predicts mean waiting room time by fitting a probability distribution given data by hospital on the proportion of patients seen in 4 hours.
  • Adjusts wait time estimate using time of day, weekday/weekend, seasonality, and location effects. Ranks hospitals by total time to treatment and returns alternatives

Why It’s Impressive:

  • Faster service for the patient
  • Helps spread load across hospitals by diverting to less‑busy sites
  • Better care outcomes through smarter routing
  • Simple, one‑click UX that hides complex modeling under the hood

Challenges We Ran Into:

  • We had only the proportion of people seen within 4 hours; to recover mean wait times, we fitted parametric probability distributions to this data, using a published paper’s curve as a reference and performing curve fitting
  • We didn’t get Claude tokens initially, so we had to go through a process to obtain them Geolocation accuracy: improved results by using Google Maps APIs
  • NHS APIs are poorly documented; much of the info is about older APIs, so we used trial and error
  • Vet distance: initially used Google Maps for all, then added a heuristic to reduce API calls
  • Finding data on how time of day affects wait times required additional research
  • Data discovery took significant time

Learning + Accomplishments:

We’ve built a lot in our website, but especially we’re proud of:

  • One‑click experience with no onboarding
  • Dynamic rankings that combine travel + wait in real time
  • A statistical model that adds predictive value without adding UX complexity

What we learned along the way:

  • Even “simple” UX can hide serious technical depth
  • Data quality and naming consistency are as hard as the modeling
  • The best hackathon projects are usable in under a minute

What’s next:

  • Collaborate with the NHS to provide live feeds which can improve wait time estimate
  • Validate predictions with hospital‑level outcomes, improving national hospital safety
  • Improve coverage to across the British isles

Category Fit:

Deliberately simple: 1 click, no login, no settings, no tutorial. All the nuance is under the hood so anyone can use it immediately.

Share this project:

Updates