An Unsupervised Approach to Achieve Supervised-Level Explainability in Healthcare Records

TL;DR

This paper introduces an unsupervised method for generating plausible and faithful explanations in healthcare record processing, eliminating the need for costly human annotations while maintaining high explanation quality.

Contribution

It presents a novel adversarial robustness training approach and the AttInGrad explanation method, achieving supervised-level explainability without supervision.

Findings

Adversarial training improves explanation plausibility.

AttInGrad outperforms previous explanation methods.

Unsupervised explanations match or surpass supervised ones.

Abstract

Electronic healthcare records are vital for patient safety as they document conditions, plans, and procedures in both free text and medical codes. Language models have significantly enhanced the processing of such records, streamlining workflows and reducing manual data entry, thereby saving healthcare providers significant resources. However, the black-box nature of these models often leaves healthcare professionals hesitant to trust them. State-of-the-art explainability methods increase model transparency but rely on human-annotated evidence spans, which are costly. In this study, we propose an approach to produce plausible and faithful explanations without needing such annotations. We demonstrate on the automated medical coding task that adversarial robustness training improves explanation plausibility and introduce AttInGrad, a new explanation method superior to previous ones. By combining both contributions in a fully unsupervised setup, we produce explanations of comparable quality, or better, to that of a supervised approach. We release our code and model weights.