Grounded Recurrent Neural Networks
This addresses the challenge of multi-label prediction in healthcare for extracting diagnoses and procedures from discharge summaries, representing an incremental improvement with a novel method for a known bottleneck.
The authors tackled the problem of extracting multiple medical concepts from clinical text by introducing the Grounded Recurrent Neural Network (GRNN), which ties labels to specific hidden state dimensions, and demonstrated its advantage over strong baselines on an ICU dataset.
In this work, we present the Grounded Recurrent Neural Network (GRNN), a recurrent neural network architecture for multi-label prediction which explicitly ties labels to specific dimensions of the recurrent hidden state (we call this process "grounding"). The approach is particularly well-suited for extracting large numbers of concepts from text. We apply the new model to address an important problem in healthcare of understanding what medical concepts are discussed in clinical text. Using a publicly available dataset derived from Intensive Care Units, we learn to label a patient's diagnoses and procedures from their discharge summary. Our evaluation shows a clear advantage to using our proposed architecture over a variety of strong baselines.