Zero-Shot Clinical Acronym Expansion via Latent Meaning Cells
This work addresses the challenge of acronym expansion in clinical text, which is crucial for healthcare professionals, but it appears incremental as it builds on existing contextual representation methods with metadata integration.
The paper tackled the problem of zero-shot clinical acronym expansion by introducing Latent Meaning Cells, a deep latent variable model that learns contextualized word representations using local lexical context and metadata, and demonstrated significant performance improvements over baselines across three datasets at a fraction of the pre-training cost.
We introduce Latent Meaning Cells, a deep latent variable model which learns contextualized representations of words by combining local lexical context and metadata. Metadata can refer to granular context, such as section type, or to more global context, such as unique document ids. Reliance on metadata for contextualized representation learning is apropos in the clinical domain where text is semi-structured and expresses high variation in topics. We evaluate the LMC model on the task of zero-shot clinical acronym expansion across three datasets. The LMC significantly outperforms a diverse set of baselines at a fraction of the pre-training cost and learns clinically coherent representations. We demonstrate that not only is metadata itself very helpful for the task, but that the LMC inference algorithm provides an additional large benefit.