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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Although recent advances in scaling large language models (LLMs) haveresulted in improvements on many NLP tasks, it remains unclear whether thesemodels trained primarily with general web text are the right tool in highlyspecialized, safety critical domains such as clinical text. Recent results havesuggested that LLMs encode a surprising amount of medical knowledge. Thisraises an important question regarding the utility of smaller domain-specificlanguage models. With the success of general-domain LLMs, is there still a needfor specialized clinical models? To investigate this question, we conduct anextensive empirical analysis of 12 language models, ranging from 220M to 175Bparameters, measuring their performance on 3 different clinical tasks that testtheir ability to parse and reason over electronic health records. As part ofour experiments, we train T5-Base and T5-Large models from scratch on clinicalnotes from MIMIC III and IV to directly investigate the efficiency of clinicaltokens. We show that relatively small specialized clinical models substantiallyoutperform all in-context learning approaches, even when finetuned on limitedannotated data. Further, we find that pretraining on clinical tokens allows forsmaller, more parameter-efficient models that either match or outperform muchlarger language models trained on general text. We release the code and themodels used under the PhysioNet Credentialed Health Data license and data useagreement.

Do We Still Need Clinical Language Models?