Inspiration

Patient follow-ups after hospital discharge are essential for continuity of care, but patients may forget appointments or struggle to organize their discharge instructions. Healthcare staff may spend considerable time reviewing documents, entering information, and tracking missed appointments. We wanted to explore how AI and workflow automation could simplify these tasks, reduce manual effort, and help healthcare teams coordinate post-discharge care more efficiently. This inspired us to develop CuraLoop, a centralized platform connecting hospital staff and patients.

What it does

CuraLoop is an AI-assisted patient follow-up and care coordination platform designed to organize post-discharge healthcare workflows. It uses AI to extract relevant follow-up dates, appointment details, and clinician-provided instructions from uploaded discharge documents. Authorized healthcare staff can review and approve the extracted information before it is used. The platform helps manage follow-up appointments, send patient reminders, track appointment status, and identify overdue follow-ups using predefined rules. A patient-facing interface allows patients to view approved instructions, check upcoming appointments, and receive reminders.

How we built it

CuraLoop is planned as a web application combining a user interface, backend services, database storage, and AI-assisted document processing. React will be used to develop the staff dashboard and patient interface, while Java Spring Boot will handle REST APIs, application logic, appointment management, and follow-up workflows. PostgreSQL or Supabase will provide database storage, and an LLM API will assist in extracting structured information from uploaded discharge documents. An email or simulated messaging service will support reminders and alerts. Authentication and role-based access control will help restrict access to sensitive information. The proposed workflow includes document upload, AI extraction, staff verification, record storage, appointment scheduling, and reminder delivery.

Challenges we ran into

Developing CuraLoop involves several important challenges, including extracting accurate information from discharge documents with different layouts, handling incomplete or ambiguous details, and ensuring that AI-generated information is verified before use. Protecting sensitive patient information and restricting access to authorized users are also essential. Another challenge is coordinating appointment records, reminder delivery, and overdue follow-up alerts reliably. To address these concerns, the proposed design includes human verification, backend validation rules, access controls, and testing with synthetic patient data.

Accomplishments that we're proud of

CuraLoop brings AI-assisted document extraction, staff verification, appointment management, patient reminders, and overdue follow-up tracking together in one proposed healthcare platform. Its main strength is its focus on a specific administrative problem rather than functioning as a generic healthcare chatbot. The design separates AI-based document processing from rule-based scheduling and appointment tracking, making the workflow more practical and easier to validate. We have also defined distinct staff and patient interfaces and a focused prototype scope suitable for a 24-hour hackathon.

What we learned

CuraLoop highlights how AI can help transform unstructured documents into organized information, while traditional programming is better suited to predictable tasks such as appointment scheduling, status tracking, and overdue detection. We also recognized the importance of human verification when handling healthcare instructions and the need to consider privacy, accuracy, accessibility, and usability from the beginning. A successful healthcare technology solution should use AI where it provides genuine value rather than adding AI features without a clear purpose.

What's next for CuraLoop

Our next step is to develop a working prototype with a staff dashboard, document upload, AI-assisted extraction, appointment management, and a patient-facing interface. We plan to test document extraction against sample discharge documents with verified expected outputs, implement reminders and overdue alerts, and evaluate performance using extraction accuracy, processing time, notification success, and overdue detection. Further improvements may include stronger security controls, multilingual reminders, accessibility enhancements, and authorized integration with hospital information systems. Our long-term goal is to make post-discharge follow-up coordination more organized, reliable, and accessible while keeping healthcare professionals responsible for approving care-related information.