From: Comparison of different feature extraction methods for applicable automated ICD coding
Feature extraction & classifiers | Macro-F1 (%) | Micro-F1 (%) | Macro-AUC (%) | Micro-AUC (%) |
---|---|---|---|---|
BoW | ||||
 LR_uni | 5.96 | 13.81 | 51.81 | 53.72 |
 SVM_uni | 24.06 | 39.13 | 58.61 | 62.61 |
 LR_uni_bi | 2.42 | 6.29 | 50.70 | 51.62 |
 SVM_uni_bi | 12.79 | 23.56 | 54.39 | 56.75 |
 LR_uni_bi_tri | 1.55 | 4.04 | 50.43 | 51.03 |
 SVM_uni_bi_tri | 8.14 | 14.76 | 52.70 | 54.00 |
W2V | ||||
 LR_word | 0.57 | 2.31 | 50.15 | 50.57 |
 SVM_word | 0.00 | 0.00 | 50.00 | 50.00 |
BERT_embeddings | ||||
 LR_char | 15.81 | 22.77 | 55.42 | 57.28 |
 SVM_char | 15.34 | 21.39 | 56.00 | 57.55 |
 LR_comb | 17.71 | 26.45 | 56.25 | 58.78 |
 SVM_comb | 18.24 | 25.75 | 57.43 | 59.68 |
BERT_finetune | ||||
 top_layer | 0.01 | 0.04 | 52.70 | 65.02 |
 whole | 1.72 | 6.71 | 68.40 | 74.87 |