CareThread

• 

  A linked report retains its explicit partial-report wording.

• 

  An older report has a matching test name but conflicting episode, order and dates.

• 

  An exported question retains its source citation and snapshot qualifications.

• 

  The actual landing page explains CareThread before entering the workspace.

Inspiration

Preparing for a clinic visit can involve sorting test orders, separate reports, and older documents that look similar. Finding a report does not explain whether it belongs to the current order or whether its source describes it as partial. CareThread began with a practical question: how could an adult patient, or a caregiver they choose, turn that document-checking work into clearer questions for the clinic?

What it does

CareThread is a working local prototype for one person's test documents from one care episode. Users add sources, inspect extracted details, review possible report associations, and explicitly decide which reports to link. Suggestions show their evidence and conflicts; missing context stays unknown.

An explicitly partial report stays partial after linking. A test with no linked report remains an unresolved document question, without implying that care was missed. Users can select and edit questions, review their supporting sources, and export a dated PDF or text sheet to take to a visit. Evidence changes flag affected questions for review while preserving edited wording.

How we built it

The interface uses React, TypeScript, and Vite. PDF.js reads supported selectable-text PDFs; conservative deterministic rules extract explicit English administrative labels and simple tables. Text input and manual details support the review workflow. Scans and photos can be retained for manual reference, but the prototype does not perform OCR.

The browser holds the current draft, while an ordered save queue writes the episode to IndexedDB through idb. PDF-lib produces question sheets with available source references. Imported-document processing runs in the browser. The current workspace supports one operator and one active tab at a time.

Challenges we ran into

The hardest challenge was keeping uncertainty visible throughout the workflow. A matching test name alone cannot establish a connection. An older report may conflict with the current order, and linking a partial report must preserve its qualification.

Another challenge was keeping saved data, the open draft, and exported questions understandable when evidence changes or storage fails. The interface distinguishes successful saves, offers retry and export, and treats each downloaded sheet as a dated snapshot.

Accomplishments that we're proud of

The prototype completes the path from document intake to explicit link review, editable questions, export, and reopening a saved episode. Its fictional demo includes a conflicting older report, a partial report, and a report that arrives later.

The reviewed release passed 70 unit tests and 24 browser tests using known synthetic scenarios in local Chromium.

What we learned

Document organization needs more than a list of files. Source references, visible conflicts, and review decisions need to travel with the questions they inform. Software checks establish specific behavior; understanding whether this workflow improves preparation requires an intended-user study, which has not yet been conducted.

What's next for CareThread

The next step is a small evaluation using the same fictional document packet in CareThread and a folder-plus-checklist workflow. The proposed comparison measures preparation time, corrections, and understanding of sources. Those findings would guide clearer document review, question wording, and recovery behavior before expanding the prototype's scope.

Built With

• idb
• indexeddb
• lucide
• manrope
• pdf-lib
• pdf.js
• playwright
• react
• typescript
• vite
• vitest