Rationale production to support clinical decision-making

TL;DR

This paper evaluates interpretability methods for neural networks predicting hospital readmission from clinical notes, highlighting the importance of domain expertise and pretraining for effective explanations.

Contribution

It compares extractive rationale methods with attention-based models in clinical NLP, emphasizing the role of domain knowledge and pretraining in interpretability.

Findings

InfoCal produces meaningful extractive rationales.

Pretraining on clinical data improves model performance.

Domain expertise enhances interpretability and trust.

Abstract

The development of neural networks for clinical artificial intelligence (AI) is reliant on interpretability, transparency, and performance. The need to delve into the black-box neural network and derive interpretable explanations of model output is paramount. A task of high clinical importance is predicting the likelihood of a patient being readmitted to hospital in the near future to enable efficient triage. With the increasing adoption of electronic health records (EHRs), there is great interest in applications of natural language processing (NLP) to clinical free-text contained within EHRs. In this work, we apply InfoCal, the current state-of-the-art model that produces extractive rationales for its predictions, to the task of predicting hospital readmission using hospital discharge notes. We compare extractive rationales produced by InfoCal to competitive transformer-based models pretrained on clinical text data and for which the attention mechanism can be used for interpretation. We find each presented model with selected interpretability or feature importance methods yield varying results, with clinical language domain expertise and pretraining critical to performance and subsequent interpretability.