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

Labeling training datasets has become a key barrier to building medicalmachine learning models. One strategy is to generate training labelsprogrammatically, for example by applying natural language processing pipelinesto text reports associated with imaging studies. We propose cross-modal dataprogramming, which generalizes this intuitive strategy in atheoretically-grounded way that enables simpler, clinician-driven input,reduces required labeling time, and improves with additional unlabeled data. Inthis approach, clinicians generate training labels for models defined over atarget modality (e.g. images or time series) by writing rules over an auxiliarymodality (e.g. text reports). The resulting technical challenge consists ofestimating the accuracies and correlations of these rules; we extend a recentunsupervised generative modeling technique to handle this cross-modal settingin a provably consistent way. Across four applications in radiography, computedtomography, and electroencephalography, and using only several hours ofclinician time, our approach matches or exceeds the efficacy ofphysician-months of hand-labeling with statistical significance, demonstratinga fundamentally faster and more flexible way of building machine learningmodels in medicine.

Cross-Modal Data Programming Enables Rapid Medical Machine Learning