Improved de-identification of physician notes through integrative modeling of both public and private medical text

Table 3 Classifier validation results

Classifier validation	Precision	F1	F10	Recall
Baseline	50	67	98	99
Boosted	59	74	98	99
Boosted + FP filtering	61	75	98	98

Validation was performed by replacing the i2b2 surrogate names with real names from Medicare and the US patent office. This was done to ensure that the model is not limited to the i2b2 surrogates. The results are similar to the original results.

ISSN: 1472-6947