Leveraging Deep Representations of Radiology Reports in Survival Analysis for Predicting Heart Failure Patient Mortality
This work addresses the problem of predicting mortality for heart failure patients using clinical texts, offering a method that reduces reliance on large annotated datasets, though it is incremental in applying BERT to this specific domain.
The paper tackled the challenge of using unstructured clinical texts in survival analysis by employing BERT-based hidden layer representations as covariates in proportional hazards models to predict heart failure patient mortality, resulting in a 5.7% average improvement in C-index and time-dependent AUC over the baseline.
Utilizing clinical texts in survival analysis is difficult because they are largely unstructured. Current automatic extraction models fail to capture textual information comprehensively since their labels are limited in scope. Furthermore, they typically require a large amount of data and high-quality expert annotations for training. In this work, we present a novel method of using BERT-based hidden layer representations of clinical texts as covariates for proportional hazards models to predict patient survival outcomes. We show that hidden layers yield notably more accurate predictions than predefined features, outperforming the previous baseline model by 5.7% on average across C-index and time-dependent AUC. We make our work publicly available at https://github.com/bionlplab/heart_failure_mortality.