Search

Full-text search

Asclepius uses SQLite FTS5 for ranked full-text search across:

• OCR text, the raw text extracted from documents
• Raw extraction data, the JSON output from LLM extraction

Usage

1. Go to Search in the sidebar
2. Type your search query
3. Results are ranked by relevance and show matching document metadata

How it works

The search uses a fuzzy LIKE query across multiple fields, searching:

• OCR text content
• Document summaries
• Doctor and facility names
• Document types
• Tags and notes

Results show:

• Document title / filename
• Document type and date
• Patient name
• Matching text snippet (highlighted)
• Link to the full document detail page

Search tips

• Search is case-insensitive
• Specific medical terms produce more precise results (e.g. “hemoglobin” rather than “blood test”)
• Patient context is applied automatically when a patient is selected in the sidebar
• With no patient selected, results include every patient you have access to

FTS5 index

The FTS5 virtual table documents_fts indexes:

• ocr_text — full OCR output
• raw_extraction — JSON extraction results

The index is kept in sync automatically via database triggers on insert, update, and delete operations on the documents table.