Working with sensitive healthcare data

Automatically redact sensitive information, ensuring privacy and regulatory adherence

Challenge

Working with sensitive healthcare data

Electronic Health Records (EHRs) are rich with personal health information, which includes sensitive patient data. Protecting patient privacy while utilizing this data for analytics or research is a major challenge. Directly accessing these notes without redaction or substitution of Personal Identifiable Information (PII) risks privacy breaches and non-compliance with regulations like HIPAA.

Solution

Redacting datasets with OmniAI

OmniAI offers a robust solution for redacting sensitive information. In this example, we'll be pulling data from physician notes in the Epic EHR system to automatically detect and redact PII, transforming names, addresses, and other identifiers into blanks or substitute names. For redactions, text we have two options:

Full Redaction
"Tyler Maran" becomes "_____ _____". For applications where PII values do not need to be displayed.
Substitution
Tyler Maran" becomes "John Doe", and a key-value store is maintained so names can be mapped back later.

Get started in a few lines of code

A schema is defined to categorize various types of PII in the physician notes, setting the stage for precise redaction or substitution.

1patient_name:
2  description: "Redacted name"
3  type: "String"
4
5physician_notes_redacted:
6  description: "Redacted version of physician_notes"
7  tool: "REDACT"
8  fields:
9    - "NAME"
10    - "ADDRESS"
11    - "SOCIAL_SECURITY"
12    - "PHONE"
13  type: "String"

Battle tested tools for top use cases

OmniAI processes the physician notes, applying the chosen redaction method. The resulting data is then structured into a new table with columns for original and redacted information, making it secure and easy to analyze.

Insights

Empower healthcare organizations to leverage their data responsibly

Post-processing, users have a safe, anonymized dataset at their disposal. This allows for the utilization of sensitive healthcare data in research and analytics without compromising patient privacy. The redacted data sets pave the way for insights into healthcare trends, treatment outcomes, and operational efficiency, all while upholding the highest standards of data privacy.