Rationale production to support clinical decision-making
This work addresses the need for interpretable AI in clinical decision-making, specifically for hospital readmission prediction, but it is incremental as it applies existing methods to a specific domain without major innovations.
The study tackled the problem of predicting hospital readmission using clinical free-text from electronic health records by applying InfoCal, a state-of-the-art model that produces extractive rationales, and compared it to transformer-based models with attention mechanisms. The results showed that performance and interpretability varied depending on clinical language domain expertise and pretraining, but no concrete numbers were provided.
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.