Skip to main content
. 2022 Jul;11(7):1216–1233. doi: 10.21037/tp-22-275

Table 5. The measurements of the method that used the TextCNN model only and the method that used our MSCNN for the 12 disease categories.

Disease Precision Recall F1-score
TextCNN MSCNN TextCNN MSCNN TextCNN MSCNN
AURI 0.56 0.75 0.59 0.77 0.58 0.76
Bronchitis 0.49 0.68 0.68 0.82 0.55 0.74
Asthma 0.48 0.68 0.26 0.45 0.32 0.5
Pharyngitis 0.41 0.58 0.25 0.42 0.29 0.48
Pneumonia 0.46 0.65 0.23 0.45 0.26 0.53
Rhinitis 0.37 0.58 0.23 0.43 0.33 0.49
Tonsillitis 0.27 0.45 0.11 0.29 0.17 0.35
Laryngitis 0.50 0.68 0.38 0.57 0.38 0.6
Nasosinusitis 0.49 0.68 0.28 0.45 0.34 0.51
FLU 0.19 0.36 0.05 0.19 0.07 0.26
FBAO 0.53 0.72 0.16 0.33 0.27 0.4
Others 0.39 0.56 0.22 0.48 0.30 0.48

Metrics included precision, recall, and F1-score. TextCNN, text convolutional neural network; MSCNN, medical-semantic-aware convolution neural network; AURI, acute upper respiratory infection; FLU, influenza; FBAO, foreign body airway obstruction.