B-CON AI · seeing · Vision and document understanding
B-CON Eyes
Reads prescriptions, reports and cards from a phone photo.
A vision model that reads what patients bring in: handwritten prescriptions, lab reports from other labs, old discharge summaries and ID cards. It turns a photo into structured data the EHR can use, so a patient's history does not stay trapped on paper.
Capabilities
What it does.
- 01Reads handwritten and printed prescriptions, including Nepali
- 02Extracts medicines, doses and durations into the medication list
- 03Reads outside lab reports into LOINC-coded results
- 04Reads citizenship, insurance and HIB cards at registration
- 05Flags low-confidence fields for a human to confirm
Specification
How it is built and served.
- Inputs
- Phone photos, scans and PDFs
- Outputs
- Structured FHIR resources with per-field confidence
- Languages
- English and Nepali, printed and handwritten
- Serving
- Self-hosted GPU alongside B-CON Brain
- Privacy
- Images processed in memory and not retained by the model service
The problem it solves
Most of a Nepali patient's history is a folded piece of paper.
Patients carry prescriptions and reports from hospital to hospital. B-CON Eyes lets them photograph it in the app and have it become part of their record, so the next doctor starts with the full picture.
The other senses