Skip to main content
. 2013 Jun 7;20(5):962–968. doi: 10.1136/amiajnl-2013-001756

Table 3.

Algorithm performance: free text and non-standard drop list allergy entries only

All allergies Drug allergies Food allergies All unique strings only
Dataset Dataset Dataset Dataset
Measure Training Testing Training Testing Training Testing Training Testing
# evaluated 8053 8958 6121 6044 1120 1888 2447 2570
Accuracy (%) 94.26 88.84 96.03 91.62 97.40 93.38 87.32 74.05
Specificity (%) 95.93 97.78 90.32 85.00 95.93 97.78 96.43 97.64
Precision (%) 99.93 99.94 99.96 99.96 99.93 99.94 99.85 99.83
Recall (%) 94.23 88.61 96.05 91.64 94.23 88.61 86.98 72.82
F-measure (%) 97.00 93.93 97.94 95.60 97.00 93.93 96.43 97.64

F-measure=harmonic mean of precision and recall.