Inspiration

Community health workers often have only a short time to understand why someone needs care. Important details can get lost between a patient’s first description and the care team. We wanted to make that first conversation clearer and easier to hand off.

What it does

CareBridge is a prototype that organizes a patient’s selected symptoms, onset, and own words into a concise handoff. It surfaces a small set of possible emergency warning signals and explains which selection triggered each prompt. A clinician remains responsible for reviewing the information and deciding what to do.

How we built it

We built a responsive, browser-based demo with HTML, CSS, and JavaScript. Its rules run locally in the browser; the prototype has no backend or AI model. The fixed rules make the prompts easy to inspect and explain, while the handoff can be copied for a care team.

Challenges we ran into

Healthcare tools need to be useful without implying that they can diagnose or rule out danger. We had to keep the prompts limited and transparent, make clear that the prototype is not clinically validated, and avoid embedding one country’s emergency number into a project intended for a global audience.

Accomplishments that we're proud of

We created a working, responsive concept that runs without an account or server. It keeps the patient’s own words alongside structured symptom selections, explains its prompts, and makes human review central to the workflow.

What we learned

In healthcare, a clear explanation and a careful handoff can matter as much as an AI feature. We also learned to distinguish an AI workflow concept from a trained model: CareBridge demonstrates an explainable intake workflow, but does not currently use machine learning or generative AI.

What's next for carebridge

We would co-design with patients and community health workers, then evaluate the prototype using fictional cases. Clinicians could assess handoff completeness, summary fidelity, and unnecessary or missed prompts. Before any real-world pilot, the workflow would need clinical validation, local emergency guidance, accessibility review, and appropriate privacy and governance safeguards.

Built With

Share this project:

Updates

Submission history