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

Contrastive learning has demonstrated promising performance in image and textdomains either in a self-supervised or a supervised manner. In this work, weextend the supervised contrastive learning framework to clinical riskprediction problems based on longitudinal electronic health records (EHR). Wepropose a general supervised contrastive loss $\mathcal{L}_{\text{ContrastiveCross Entropy} } + \lambda \mathcal{L}_{\text{Supervised ContrastiveRegularizer}}$ for learning both binary classification (e.g. in-hospitalmortality prediction) and multi-label classification (e.g. phenotyping) in aunified framework. Our supervised contrastive loss practices the key idea ofcontrastive learning, namely, pulling similar samples closer and pushingdissimilar ones apart from each other, simultaneously by its two components:$\mathcal{L}_{\text{Contrastive Cross Entropy} }$ tries to contrast sampleswith learned anchors which represent positive and negative clusters, and$\mathcal{L}_{\text{Supervised Contrastive Regularizer}}$ tries to contrastsamples with each other according to their supervised labels. We propose twoversions of the above supervised contrastive loss and our experiments onreal-world EHR data demonstrate that our proposed loss functions show benefitsin improving the performance of strong baselines and even state-of-the-artmodels on benchmarking tasks for clinical risk predictions. Our loss functionswork well with extremely imbalanced data which are common for clinical riskprediction problems. Our loss functions can be easily used to replace (binaryor multi-label) cross-entropy loss adopted in existing clinical predictivemodels. The Pytorch code is released at\url{https://github.com/calvin-zcx/SCEHR}.

SCEHR: Supervised Contrastive Learning for Clinical Risk Prediction using Electronic Health Records