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

Objective: The generalizability of clinical large language models is usuallyignored during the model development process. This study evaluated thegeneralizability of BERT-based clinical NLP models across different clinicalsettings through a breast cancer phenotype extraction task.  Materials and Methods: Two clinical corpora of breast cancer patients werecollected from the electronic health records from the University of Minnesotaand the Mayo Clinic, and annotated following the same guideline. We developedthree types of NLP models (i.e., conditional random field, bi-directional longshort-term memory and CancerBERT) to extract cancer phenotypes from clinicaltexts. The models were evaluated for their generalizability on different testsets with different learning strategies (model transfer vs. locally trained).The entity coverage score was assessed with their association with the modelperformances.  Results: We manually annotated 200 and 161 clinical documents at UMN and MC,respectively. The corpora of the two institutes were found to have highersimilarity between the target entities than the overall corpora. The CancerBERTmodels obtained the best performances among the independent test sets from twoclinical institutes and the permutation test set. The CancerBERT modeldeveloped in one institute and further fine-tuned in another institute achievedreasonable performance compared to the model developed on local data (micro-F1:0.925 vs 0.932).  Conclusions: The results indicate the CancerBERT model has the best learningability and generalizability among the three types of clinical NLP models. Thegeneralizability of the models was found to be correlated with the similarityof the target entities between the corpora.

A Cross-institutional Evaluation on Breast Cancer Phenotyping NLP Algorithms on Electronic Health Records