Table 1.
Summarization of the different approaches and their results. Training and test Pearson correlation coefficient scores are rounded to 3 decimal places.
Approach | Training set | Test set | |
Approach 0: baseline | |||
ClinicalBERT | 0.850 | 0.859 | |
Approach 1: voting regression | |||
Enhanced BERT | 0.851 | 0.859 | |
Voting Regression | 0.860 | 0.849a | |
Approach 2: M-Heads | |||
Enhanced BERT with M-Heads | 0.853 | 0.876a,b | |
Enhanced BERT with M-Heads + Med. Graph | 0.853 | 0.883c | |
Approach 3: medication graph | |||
Voting Regression + Med. Graph | 0.862 | 0.862a |
aOur submissions.
bWe submitted a score of 0.869 for this setting because we were able to use only 10 k-folds due to a shortage of time.
cOur best result of the test set.