OCR to JSON & SQLite

I need a clean, end-to-end pipeline that pulls text from mixed printed-and-handwritten patient assessment forms, turns that text into well-structured JSON, and then inserts each record into a SQLite table.

Key points you should know before diving in:
• Source material is mostly the same template but there are occasional layout quirks, so the solution has to be tolerant of minor shifts or missing fields.
• Tesseract will be the OCR engine. I already use basic grayscale and thresholding—feel free to add any preprocessing steps that boost accuracy on handwriting without slowing the run time too much.
• The only compulsory fields for this phase are Patient information (name, DOB, ID, etc.) and Medical history sections. If you structure the parser cleanly, we can bolt on Assessment results later.
• Once text is captured, map it to a clear JSON schema and push each JSON object straight into a SQLite database. A simple, version-controlled schema migration script is expected.

Deliverables
1. Source code (Python preferred) with reproducible environment file.
2. Sample dataset showing raw scan → JSON → SQLite row for at least five test forms.
3. README that covers setup, command syntax, and a quick note on how to tweak field mappings for future form variants.

I already have a starter repo; you are free to refactor or rebuild as long as the final workflow runs in one command. Let me know the libraries you plan to use beyond Tesseract so I can verify deployment compatibility.