English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Long-form clinical summarization of hospital admissions has real-worldsignificance because of its potential to help both clinicians and patients. Thefaithfulness of summaries is critical to their safe usage in clinical settings.To better understand the limitations of abstractive systems, as well as thesuitability of existing evaluation metrics, we benchmark faithfulness metricsagainst fine-grained human annotations for model-generated summaries of apatient's Brief Hospital Course. We create a corpus of patient hospitaladmissions and summaries for a cohort of HIV patients, each with complexmedical histories. Annotators are presented with summaries and source notes,and asked to categorize manually highlighted summary elements (clinicalentities like conditions and medications as well as actions like "followingup") into one of three categories: ``Incorrect,'' ``Missing,'' and ``Not inNotes.'' We meta-evaluate a broad set of proposed faithfulness metrics and,across metrics, explore the importance of domain adaptation (e.g. the impact ofin-domain pre-training and metric fine-tuning), the use of source-summaryalignments, and the effects of distilling a single metric from an ensemble ofpre-existing metrics. Off-the-shelf metrics with no exposure to clinical textcorrelate well yet overly rely on summary extractiveness. As a practical guideto long-form clinical narrative summarization, we find that most metricscorrelate best to human judgments when provided with one summary sentence at atime and a minimal set of relevant source context.

A Meta-Evaluation of Faithfulness Metrics for Long-Form Hospital-Course Summarization