Skip to main content

Table 9 Result for the cardiovascular and skin-specific models on in- and out-of-domain test data sets compared to the model trained on all of the data

From: Classifying patient and professional voice in social media health posts

Model

Other: F1

Patient voice: F1

Prof. Voice: F1

Macro F1

Acc.

Test

Cardio

0.94

0.87

0.37

0.73

0.91

Cardio:

Skin

0.88

0.73

0.06

0.56

0.83

2791

All

0.93

0.86

0.21

0.67

0.90

 

Cardio

0.69

0.63

0.07

0.46

0.66

Skin:

Skin

0.71

0.82

0.34

0.62

0.77

3083

All

0.73

0.83

0.16

0.57

0.76

 
  1. We report F1 scores per label, macro-average F1 and accuracy across all three labels as well as the size of the test set