Skip to main content

Table 1 Information extraction accuracy

From: Identification of methicillin-resistant Staphylococcus aureus within the Nation’s Veterans Affairs Medical Centers using natural language processing

Training Set

     
 

Records Reviewed : 62,500 (53,627 records from SLC only annotated for MRSA)

   

Sensitivity

Specificity

PPV

NPV

 

Staphylococcus aureus

99.6 (4026/4044)

99.9 (4828/4829)

99.9 (4026/4027)

99.6 (4828/4846)

  

Methicillin Resistance

99.9 (2789/2790)

99.9 (59701/59710)

99.7 (2786/2795)

99.9 (59701/59705)

Validation Set

     
 

Electronic Records Reviewed: 5,927

   
   

Sensitivity

Specificity

PPV

NPV

 

Staphylococcus aureus

100 (2739/2739)

99.9 (3185/3188)

99.9 (2739/2742)

100 (3185/3185)

  

Methicillin Resistance

100 (1460/1460)

99.9 (4465/4467)

99.9 (1460/1462)

100 (4465/4465)

 

Expert Reviewed Records: 3,092

   
   

Sensitivity

Specificity

PPV

NPV

 

Staphylococcus aureus

98.3 (1348/1372)

99.7 (1714/1720)

99.6 (1348/1354)

98.6 (1714/1738)

  

Methicillin Resistance

99.2 (703/710)

99.4 (2368/2383)

97.9 (703/718)

99.8 (2368/2374)

  1. PPV - positive predictive value, NPV - negative predictive value depicts the accuracy of the extraction process on the training set (both electronic and expert-reviewed data sets combined), as well as on the validation set (reported separately). Both numbers and percentages are supplied.