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/

Zendy Plus

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

Zendy Tools

Zendy Open

Plain language summaries (PLSs) are essential for facilitating effectivecommunication between clinicians and patients by making complex medicalinformation easier for laypeople to understand and act upon. Large languagemodels (LLMs) have recently shown promise in automating PLS generation, buttheir effectiveness in supporting health information comprehension remainsunclear. Prior evaluations have generally relied on automated scores that donot measure understandability directly, or subjective Likert-scale ratings fromconvenience samples with limited generalizability. To address these gaps, weconducted a large-scale crowdsourced evaluation of LLM-generated PLSs usingAmazon Mechanical Turk with 150 participants. We assessed PLS quality throughsubjective Likert-scale ratings focusing on simplicity, informativeness,coherence, and faithfulness; and objective multiple-choice comprehension andrecall measures of reader understanding. Additionally, we examined thealignment between 10 automated evaluation metrics and human judgments. Ourfindings indicate that while LLMs can generate PLSs that appearindistinguishable from human-written ones in subjective evaluations,human-written PLSs lead to significantly better comprehension. Furthermore,automated evaluation metrics fail to reflect human judgment, calling intoquestion their suitability for evaluating PLSs. This is the first study tosystematically evaluate LLM-generated PLSs based on both reader preferences andcomprehension outcomes. Our findings highlight the need for evaluationframeworks that move beyond surface-level quality and for generation methodsthat explicitly optimize for layperson comprehension.

Are LLM-generated plain language summaries truly understandable? A large-scale crowdsourced evaluation