Radiologists' memory as a data protection risk: a worst-case stress test for chest radiograph re-identification.

Area of Science:

• Medical imaging and radiology
• Health informatics
• Patient privacy

Background:

• Re-identification risk is a concern for digital health records.
• Radiologists' memory may contribute to re-identification.
• Understanding this risk is crucial for data security.

Purpose of the Study:

• To quantify the re-identification risk associated with radiologists' memory of chest radiographs.
• To assess how often radiologists can recall explicit identifiers from prior readings.
• To determine the reliability of memory-based recognition in a controlled setting.

Main Methods:

• Prospective, multicenter, web-based reader study with 33 radiologists.
• Two reading phases: initial interpretation and follow-up readings after a 24-hour interval.
• Radiologists indicated patient recall and attempted to recall explicit identifiers (pseudonym, case position).

Main Results:

• Radiologists reported remembering patients in 35.1% of examinations.
• Sensitivity for recognizing repeated targets was 50.0%, but 20.2% of new cases were falsely recalled.
• Correct recall of explicit identifiers occurred in only 1.3% of examinations.

Conclusions:

• Radiologists' memory is a limited factor in re-identification risk.
• Familiarity does not reliably translate into accurate identification.
• False-positive recognition is common, highlighting the need for robust data protection measures beyond memory assessment.