U.S. flag

An official website of the United States government

Talk to the Veterans Crisis Line now
Skip to main content

Last updated

Optical character recognition (OCR) followed by natural language processing (NLP) on scanned documents to expedite clinical care

Share Print

Clinical care is frequently delayed because of the time it takes for staff to read and process outside medical records. We seek to use optical character recognition followed by natural language processing to classify documents and extract important clinical information from documents as they come in. This will allow us to streamline administrative processes and expedite clinical care.

This innovation is emerging and worth watching as it is being assessed in early implementations. See more emerging innovations.

Adoptions:

1 in-progress

Contact Team

Overview

Problem

Outside medical records enter our system as non-searchable PDF documents primarily. These are difficult to triage in an automated fashion and staffing shortages make it difficult to process them in a timely fashion. This can result in excessive staffing needs, delays in clinical care, and gaps in care.

Solution

A vendor that already has a contract with VBA has performed optical character recognition (OCR) and natural language processing (NLP) to classify these documents and extract relevant data. We are working with them to streamline the process and optimize the clinical triage.

Results

The initial proof of concept showed accuracy >90%.

Diffusion tracker

Does not include Clinical Resource Hubs (CRH)

Statuses

Implementation

Timeline

  • 6 months to go live
    Model training, data connections, EMR integration

Core Resources

Resource type Resource description
PEOPLE
  • VistA integration specialist
  • Project manager
  • Clinical lead
PROCESSES
  • CPRS/VistA integration
TOOLS
  • OCR and NLP Software
  • Cloud computing instance

Contact

Comment

Comments and replies are disabled for retired innovations and non-VA users.

VA User (Chief of Business Intelligence) posted

Is it possible to utilize Microsoft Power Automate for OCR as it is readily available to VHA without having to use a vendor?

Email

Email with questions about this innovation.

About

Origin story

After seeing issues around the processing of both mailed documents and faxed documents, Dr. Blake Anderson initiated this process.
BESbswy