Improving Mortality Prediction using Text Embeddings
Do text embeddings of diagnosis generated by BERT carry any clinical benefits in low-resource settings?
Do text embeddings of diagnosis generated by BERT carry any clinical benefits in low-resource settings?
How can we customize our embedding space by incorporating output class structure?