Table 4 Performance evaluation of new models.

From: A patient-centered approach to developing and validating a natural language processing model for extracting patient-reported symptoms

Methods

Attributes

BERT-CRF

LUKE

MedNER-CR-JA (PCA)

Precision

Recall

F1

Precision

Recall

F1

Precision

Recall

F1

Exact match

All

0.87

0.88

0.87

0.69

0.71

0.70

0.59

0.35

0.44

Positive

0.86

0.88

0.87

0.66

0.69

0.67

0.57

0.45

0.51

Suspicious

0.15

0.15

0.13

0.05

0.02

0.03

0.20

0.11

0.14

Negative

0.87

0.89

0.88

0.72

0.74

0.73

0.62

0.30

0.40

General

0.60

0.32

0.40

0.21

0.23

0.20

0.16

0.13

0.14

Position match

–

0.89

0.90

0.90

0.72

0.73

0.73

0.69

0.42

0.52

Back to article page