Analyzer Profiles
De-ID offers three analyzer profiles to match different de-identification needs and regulatory requirements. Below are the general guidelines for flagging potential identifiers, otherwise known as Personally Identifiable Information.
Minimal Profile
Targets the most sensitive identifiers essential for basic compliance:
- Social Security Numbers
- Names
- Phone Numbers
- Email Addresses
- Rare Diseases
- Network IDs (IP and MAC addresses)
Standard Profile
Includes all Minimal categories plus additional commonly identifying information:
- URLs
- Organizations
- Locations
- Dates
Aggressive Profile
Provides comprehensive de-identification including all Standard categories plus:
- LGBTQI+ identities
- Age references
- Numbers with timeframes
- Other numerical identifiers (driver's licenses, passports, medical IDs)
- Miscellaneous potentially identifying information
All profiles maintain confidence levels of 0.9 and have been designed and tested to serve Safe Harbor and NIH standards for human subject data protection.
Best Practices
When using De-ID, it can be helpful to understand that the initial processing outcomes presented are the results from an automated two-phase approach. Following this initial processing, best practices suggest the importance of human oversight and intervention. This human review can be critical for flagging any important omissions and, most importantly, assure de-identification decisions that respect the overall sensitivity of the data and protect the fundamental purpose of the data collection itself. That is, we believe that the primary investigators of the study in which data were collected are best positioned to understand and respond to the spirit and expectations of academic integrity for the protection of their institutions, their work, and the research participants who provided the data in question.
Two-Phase De-Identification Automation
- Initial Analysis: Multiple independent algorithms scan and identify PII category information simultaneously
- Reconciliation Phase: A hierarchical decision-tree process resolves conflicts and overlaps, based on a set of rules prioritizing more complete PII strings and maximizing the potential utility of the data post processing.
Human Oversight: While De-ID provides powerful automation, the Expert Determination method requires human review. Particularly as you become familiar and comfortable with how De-ID automation operations, we strongly recommend conducing a complete file review beyond the flagged items, as potentially identifying information may exist that wasn't automatically detected.
Color-Coded Guidance: The system uses visual indicators to distinguish between essential replacements and discretionary modifications, helping users make informed decisions about each flagged item.