Table 5 Coding results for Fuwai dataset with \(f_s = 20\)
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 | 52.95 | 82.99 | 71.82 | 86.88 |
 SVM_uni | 47.41 | 82.70 | 70.46 | 86.73 |
 LR_uni_bi | 46.25 | 80.79 | 69.10 | 85.48 |
 SVM_uni_bi | 37.70 | 79.07 | 66.11 | 84.13 |
 LR_uni_bi_tri | 39.12 | 72.85 | 65.93 | 80.49 |
 SVM_uni_bi_tri | 27.24 | 67.62 | 61.68 | 76.77 |
W2V | ||||
 LR_word | 22.81 | 63.29 | 58.79 | 74.85 |
 SVM_word | 12.74 | 53.46 | 55.07 | 68.99 |
 LR_char | 19.16 | 58.43 | 57.17 | 72.09 |
 SVM_char | 8.16 | 45.92 | 53.19 | 65.40 |
 LR_comb | 29.08 | 69.02 | 61.32 | 78.13 |
 SVM_comb | 16.92 | 61.84 | 56.97 | 73.39 |
RoBERTa_embeddings | ||||
 LR_char | 34.75 | 69.03 | 63.89 | 79.25 |
 SVM_char | 23.41 | 64.75 | 59.58 | 75.74 |
 LR_comb | 39.44 | 74.32 | 66.00 | 82.17 |
 SVM_comb | 29.64 | 70.59 | 62.16 | 79.01 |
RoBERTa_finetune | ||||
 top_layer | 0.67 | 31.06 | 62.83 | 84.21 |
 whole | 2.43 | 41.25 | 75.00 | 90.26 |