作者: Clive J. Hoggart
DOI: 10.1101/355479
关键词:
摘要: There is increasing interest in developing point of care tests to diagnose disease and predict prognosis based upon biomarker signatures RNA or protein expression levels. Technology measure the required biomarkers accurately a time-frame useful health professionals will be easier develop by minimising number measured. In this paper we describe Parallel Regularised Regression Model Search (PReMS) method which designed estimate parsimonious prediction models. Given set potential PReMS searches over many logistic regression models constructed from optimal subsets biomarkers, iteratively model size. Zero centred Gaussian prior distributions are assigned all coefficients induce shrinkage. The estimates shrinkage parameter, for each size We apply six freely available data sets compare its performance with LASSO SCAD algorithms terms covariates model, accuracy, as measured area under receiver operator curve (AUC) root predicted mean square error, calibration. show that typically selects fewer than both but has comparable predictive accuracy.