Skip to main content

Table 5 Categorization of algorithm false negatives by PHI type on test corpus.

From: Automated de-identification of free-text medical records

PHI Type # False negatives in 296,400 words/1,836 nursing notes # False negatives per 100,000 words Recall
Full name 4 † 1  
Last name 14 † 5  
First name 31 † 11  
Location (not street address) 7 2  
Full date 2 1 Unknown
Partial date 9 3  
Year 8 3  
Age over 89 3 1  
Overall 78 27 0.94 (estimated)
  1. † None of these names were actually patient names, and therefore were non-critical PHI.