Unsupervised Pseudo-Labeling for Extractive Summarization on Electronic Health Records
This work addresses the cost and expertise barriers for physicians in managing EHRs, though it is incremental as it adapts pseudo-labeling to a specific domain.
The paper tackled the problem of expensive supervised training for extractive summarization on Electronic Health Records by generating pseudo-labels from intrinsic correlations between records, resulting in an effective model for summarizing disease-specific information without external annotation.
Extractive summarization is very useful for physicians to better manage and digest Electronic Health Records (EHRs). However, the training of a supervised model requires disease-specific medical background and is thus very expensive. We studied how to utilize the intrinsic correlation between multiple EHRs to generate pseudo-labels and train a supervised model with no external annotation. Experiments on real-patient data validate that our model is effective in summarizing crucial disease-specific information for patients.