Manually scan & parse documents/OCR

[AnalogJ]

while I do plan to eventually support scanned/OCR'd PDF file importing, its pretty low on my list. There's just no standardization there, so it'll be almost impossible to import that data in a meaningful way.

we already have a bare-bones manual-entry form where you can enter conditions/procedures/medications (and add attachments).

---

• every FHIR table has a "text" column, I'd populate that with the content extracted from the PDF using OCR - https://github.com/fastenhealth/fasten-onprem/blob/main/backend/pkg/models/database/fhir_document_reference.go#L145
• the text field is not inteded to be displayed as-is (text vs RTF/markdown), just used for searching. Is that suficient or your use-case? (ie. display the PDF/image in the UI, search against the text field, so no copy-paste).
• OCR - go supported library https://github.com/otiai10/gosseract - I've used it before, its pretty decent.
• OCR - we should put the implementation behind an interface, so we can support different OCR engines (disabled/embedded when running in desktop mode, ocr server via hostname when running on cloud/docker)

[nitantsoni]

How do I use the JSON upload option for manually adding my labs? Is there a particular format that they need to be converted to?

[AnalogJ]

@nitantsoni

Fasten only supports FHIR Bundles in JSON format at the moment. I've attached a zip file containing a FHIR Bundle from Synthea that you can use for testing.

Abraham100_Heller342_262b819a-5193-404a-9787-b7f599358035.json.zip

If your lab results are stored as Observation's within a Bundle, Fasten should be able to parse them.

[nitantsoni]

Thanks!