Background

Our HLS emerging opportunities team recently decided to look at Text Analytics for Heath (part of Azure Cognitive Services) and deliver an end-to-end MVP solution to better understand how it could work in a real-world scenario. Before building the solution, the team talked through several potential use-cases that I will outline below.

Potential Use-Cases

Patient Intake

This use-case looked to automate some of the front-desk processes of a clinic by scanning the patient intake form, using OCR to convert any hand-written data to text, running the result through Text Analytics for Health, and extracting any medical information from the result to update the patient’s Electronic Medical Record (EMR).

PII (personally identifiable information) Detection

When reviewing patient research data, the Text Analytics for Health service could be used to ensure removal of any PII.

Clinical Trials

The ability to scan paper reports from participants, allowing research teams to receive updates in near real-time.

Add diagnosis data to patient record

This would allow patients with paper test results obtained outside of their primary provider to have them added to their patient health record.

Symptom Matching

Given a set of symptoms, allow epidemiologists to determine the number of patients with similar symptoms to better determine when outbreaks occur.

Physician Coaching

Allow physicians in training to learn from the diagnosis more easily

Recommendation Explanation

Better explanation of a physician’s recommendation to patients to help make it less necessary for them to search online themselves based on notes from the discussion.

Our team’s choice

Our team chose to build upon the Patient Intake scenario described above. To make this use-case come to life, we considered building the following:

• UI form to submit data
• API
• Power Automate/Logic App/Function App triggered by scan event to send scan input to OCR Service
• Back-end for sending data to Text Analytics for Health API

You can view the results of this effort in our Fast Pass GitHub Repository!