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

Tabular datasets with low-sample-size or many variables are prevalent inbiomedicine. Practitioners in this domain prefer linear or tree-based modelsover neural networks since the latter are harder to interpret and tend tooverfit when applied to tabular datasets. To address these neural networks'shortcomings, we propose an intrinsically interpretable network forheterogeneous biomedical data. We design a locally sparse neural network wherethe local sparsity is learned to identify the subset of most relevant featuresfor each sample. This sample-specific sparsity is predicted via a\textit{gating} network, which is trained in tandem with the\textit{prediction} network. By forcing the model to select a subset of themost informative features for each sample, we reduce model overfitting inlow-sample-size data and obtain an interpretable model. We demonstrate that ourmethod outperforms state-of-the-art models when applied to synthetic orreal-world biomedical datasets using extensive experiments. Furthermore, theproposed framework dramatically outperforms existing schemes when evaluatingits interpretability capabilities. Finally, we demonstrate the applicability ofour model to two important biomedical tasks: survival analysis and marker geneidentification.

Locally Sparse Neural Networks for Tabular Biomedical Data