Modelling EHR timeseries by restricting feature interaction

TL;DR

This paper introduces a recurrent neural network model that limits feature interactions to improve the prediction of clinical outcomes from electronic health record time series, especially with missing data.

Contribution

It proposes a novel RNN architecture that reduces overfitting by restricting feature interactions, enhancing predictive accuracy on EHR data.

Findings

Improved AU-ROC by 1.1% for mortality prediction on MIMIC-III.

Achieved 1.0% and 2.2% improvements in mortality and AKI predictions.

Model outperforms existing state-of-the-art methods on key clinical prediction tasks.

Abstract

Time series data are prevalent in electronic health records, mostly in the form of physiological parameters such as vital signs and lab tests. The patterns of these values may be significant indicators of patients' clinical states and there might be patterns that are unknown to clinicians but are highly predictive of some outcomes. Many of these values are also missing which makes it difficult to apply existing methods like decision trees. We propose a recurrent neural network model that reduces overfitting to noisy observations by limiting interactions between features. We analyze its performance on mortality, ICD-9 and AKI prediction from observational values on the Medical Information Mart for Intensive Care III (MIMIC-III) dataset. Our models result in an improvement of 1.1% [p<0.01] in AU-ROC for mortality prediction under the MetaVision subset and 1.0% and 2.2% [p<0.01] respectively for mortality and AKI under the full MIMIC-III dataset compared to existing state-of-the-art interpolation, embedding and decay-based recurrent models.