Skip to main content

Showing 1–2 of 2 results for author: Leeuwenberg, A M

Searching in archive stat. Search in all archives.
.
  1. arXiv:2301.06570  [pdf

    cs.CL stat.ME

    Cross-institution text mining to uncover clinical associations: a case study relating social factors and code status in intensive care medicine

    Authors: Madhumita Sushil, Atul J. Butte, Ewoud Schuit, Maarten van Smeden, Artuur M. Leeuwenberg

    Abstract: Objective: Text mining of clinical notes embedded in electronic medical records is increasingly used to extract patient characteristics otherwise not or only partly available, to assess their association with relevant health outcomes. As manual data labeling needed to develop text mining models is resource intensive, we investigated whether off-the-shelf text mining models developed at external in… ▽ More

    Submitted 16 January, 2023; originally announced January 2023.

    MSC Class: 68T50; 68U35; 62-xx; 62P10; 92C60; 92D30 ACM Class: I.2.7; G.3

  2. arXiv:2101.01603  [pdf, other

    stat.ME

    Comparing methods addressing multi-collinearity when developing prediction models

    Authors: Artuur M. Leeuwenberg, Maarten van Smeden, Johannes A. Langendijk, Arjen van der Schaaf, Murielle E. Mauer, Karel G. M. Moons, Johannes B. Reitsma, Ewoud Schuit

    Abstract: Clinical prediction models are developed widely across medical disciplines. When predictors in such models are highly collinear, unexpected or spurious predictor-outcome associations may occur, thereby potentially reducing face-validity and explainability of the prediction model. Collinearity can be dealt with by exclusion of collinear predictors, but when there is no a priori motivation (besides… ▽ More

    Submitted 5 January, 2021; originally announced January 2021.

    MSC Class: 60 ACM Class: G.3