Description
Summary We need a lightweight, reliable Python script/tool that ingests multi-page PDF medical records, uses an LLM API (OpenAI, Claude, or Gemini) to extract structured clinical events, deterministically sorts them chronologically, and outputs a formatted Markdown/Word narrative report with page-level citations. Key Deliverables: -PDF parsing and text/OCR handling (handling digital exports and scanned records). -LLM integration using structured output (JSON schema) with prompt engineering for medical accuracy. -Python-based chronological sorting and deduplication. -Export engine generating clean .docx or .md reports. -A simple user interface (e.g., Streamlit desktop app or local command-line script) so a non-technical user can drop in a PDF and click "Generate". -Clear setup instructions and 30 days of troubleshooting support. Required Skills: -Strong Python (PDF libraries like PyMuPDF/pdfplumber, pydantic, Streamlit). -Experience with LLM APIs (Anthropic, OpenAI, or Google Cloud Vertex AI) and structured JSON outputs. -Familiarity with healthcare data handling best practices (HIPAA/security awareness).