A Neural Attention Model for Categorizing Patient Safety Events

TL;DR

This paper introduces a neural attention model designed to classify patient safety event reports, significantly improving the accuracy of understanding safety narratives and aiding in medical error prevention.

Contribution

The paper presents the first neural attention-based architecture for categorizing patient safety events from narrative reports.

Findings

Significant accuracy improvements over existing methods

Effective encoding of long safety report sequences

Validated on large-scale real-world datasets

Abstract

Medical errors are leading causes of death in the US and as such, prevention of these errors is paramount to promoting health care. Patient Safety Event reports are narratives describing potential adverse events to the patients and are important in identifying and preventing medical errors. We present a neural network architecture for identifying the type of safety events which is the first step in understanding these narratives. Our proposed model is based on a soft neural attention model to improve the effectiveness of encoding long sequences. Empirical results on two large-scale real-world datasets of patient safety reports demonstrate the effectiveness of our method with significant improvements over existing methods.