AI System for Preparing Test Documents for Patient Files

The problem

In daily operations, a document may be added to the wrong patient file. When the scanned test document, patient index, request date and document glossary are spread across different screens, documents or team records, the issue may not be visible in time.

This fragmented data delays preparation of the filing-match suggestion and forces staff to search again for supporting information. Before deciding, the medical records officer must complete missing fields and verify conflicts one by one.

A common exception is that an identity field may be missing or illegible. If this context is not checked, an incomplete or incorrect workflow could place the document in another patient's file, so its output cannot be applied directly.

What we set out to improve

• Number of match candidates rejected by the officer
• Rate of outputs corrected or rejected by the medical records officer
• Time between the source record and the review output

How the system works

1. 01 Define the review scope and the authorised medical records officer role
2. 02 Receive the scanned test document, patient index, request date and document glossary
3. 03 Check source identity, date, version and data freshness
4. 04 Extract identity, date and type fields
5. 05 Separate exact and possible matches
6. 06 Check the exception that an identity field may be missing or illegible
7. 07 Place low-confidence or conflicting records in a separate human review queue
8. 08 Write the source identity, rule, output and time to the audit trail
9. 09 Have the medical records officer approve, correct, defer or reject the output

Methods we used

• Cross-validation of OCR fields against identity and date rules
• Source, format, version and mandatory-field checks for scanned test documents, the patient index, request dates and the document glossary
• A missing-data warning instead of inference when inputs are absent or conflicting
• Separate assessment of exceptions and the main rule
• Joint retention of the source identity, rule result, filing-match suggestion and human decision

Where people stay involved

Data and security

Scanned test documents, the patient index, request dates and the document glossary are accessed with the narrowest permissions required for the role. Identity and sensitive fields are masked where possible, and only the necessary section rather than the entire raw file is sent to the model. Access, output and the medical records officer's decision are logged; production use does not begin until the organisation approves data retention and location.

Who this suits

Frequently asked questions