On the Effects of Low-Quality Training Data on Information Extraction from Clinical Reports

TL;DR

This study investigates how low-quality training data, due to differing annotators, impacts the accuracy of clinical information extraction systems, revealing that disagreements often do not significantly affect system performance.

Contribution

It provides an empirical analysis of the effects of annotation quality and coder disagreement on information extraction accuracy in clinical texts.

Findings

Disagreement between coders is substantial but often not statistically significant.

Training data annotated by different coders can still produce comparable extraction accuracy.

Annotation quality impacts vary depending on the specific extraction task.

Abstract

In the last five years there has been a flurry of work on information extraction from clinical documents, i.e., on algorithms capable of extracting, from the informal and unstructured texts that are generated during everyday clinical practice, mentions of concepts relevant to such practice. Most of this literature is about methods based on supervised learning, i.e., methods for training an information extraction system from manually annotated examples. While a lot of work has been devoted to devising learning methods that generate more and more accurate information extractors, no work has been devoted to investigating the effect of the quality of training data on the learning process. Low quality in training data often derives from the fact that the person who has annotated the data is different from the one against whose judgment the automatically annotated data must be evaluated. In this paper we test the impact of such data quality issues on the accuracy of information extraction systems as applied to the clinical domain. We do this by comparing the accuracy deriving from training data annotated by the authoritative coder (i.e., the one who has also annotated the test data, and by whose judgment we must abide), with the accuracy deriving from training data annotated by a different coder. The results indicate that, although the disagreement between the two coders (as measured on the training set) is substantial, the difference is (surprisingly enough) not always statistically significant.