CareBridge AI

Inspiration

Patient information should move safely whenever a patient moves between hospitals, wards, ICUs, or discharge. We identified a gap where important changes or missing information may not be immediately visible during a handoff.

What it does

CareBridge AI compares previous and current patient records to identify missing information, medication changes, potential contradictions, and undocumented changes.

It generates a simple Handoff Safety Card showing:

What changed
What is missing
What may conflict
What needs verification

CareBridge is designed as a safety and coordination tool, not a diagnostic or prescribing system.

How we built it

For the prototype, we plan to use Python and Streamlit with OCR and document-processing techniques to extract structured information from medical records. An AI/NLP-assisted comparison engine will compare previous and current records and generate the Handoff Safety Card.

We will use synthetic or de-identified patient records for development and testing.

Challenges we ran into

The main challenge is handling different medical document formats and identifying important information consistently. Another challenge is making the system useful without allowing AI to make clinical decisions.

Our approach is to focus on detecting changes, missing information, and potential conflicts while keeping the final decision with qualified healthcare professionals.

Accomplishments that we're proud of

We developed a clear concept for addressing information gaps during patient handoffs and designed the Handoff Fingerprint approach to summarize what changed, what disappeared, and what needs verification.

We also defined a safety-first scope where CareBridge supports clinicians instead of replacing clinical judgment.

What we learned

We learned that a healthcare AI solution should not only focus on prediction. It should also focus on reliability, explainability, privacy, and human verification.

We also learned how important information continuity is when a patient moves between different points of care.

What's next for CareBridge AI

Our next step is to build a functional prototype that can process sample medical documents and demonstrate the complete workflow:

Upload records → Extract information → Compare records → Detect changes and gaps → Generate Handoff Safety Card.

Future versions could support EHR integration, multilingual documents, emergency handoffs, longitudinal patient timelines, and authorized digital care passports.