Prediction of cancer outcome with microarrays: a multiple random validation strategy.

By S. Michiels, S. Koscielny, C. Hill. The Lancet, 2005

1,040 citations recorded by the source corpus.

Subfields: Cancer Research, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/1ef84899fbd136a17d08cbaa4baded5c4997e8d6

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cancer Research

In Molecular Biology

Citations in the corpus

Google Goes Cancer: Improving Outcome Prediction for Cancer Patients by Network-Based Ranking of Marker Genes (2012, methodology): “To get a robust estimate on the classification error rate, we adopted the multiple random validation strategy described by [22].”
Gene Expression-Based Classification of Non-Small Cell Lung Carcinomas and Survival Prediction (2010, background): “It has been noted before that there is very little, if any, overlap between the reported prognostic signatures for NSCLC [25,31].”
Algorithms for Discovery of Multiple Markov Boundaries (2013, methodology): “Description: In resampling-based methods, multiple variable sets are extracted by repeatedly applying a variable selection method to different bootstrap samples of the data (Ein-Dor et al., 2005; Michiels et al., 2005; Roepman et al., 2006; Statnikov and Aliferis, 2010a).”
A modular analysis framework for blood genomics studies: application to systemic lupus erythematosus (2008, background): “…results obtained by different laboratories and across platforms raised suspicion about the validity of these results, and this remains a major concern (Bammler et al., 2005; Frantz, 2005; Ioannidis, 2005; Irizarry et al., 2005; Larkin et al., 2005; Michiels et al., 2005; Shi et al., 2006).”
Machine learning applications in cancer prognosis and prediction (2014, background): “Various aspects regarding the prediction of cancer outcome based on gene expression signatures are discussed in [8,9].”
A review of microarray datasets and applied feature selection methods (2014, background): “For example, in [82] it is reported that reanalysis of data from the seven largest published microarray-based studies that have attempted to predict the prognosis of cancer patients reveals that five of those seven did not classify patients better than chance.”
The balance of reproducibility, sensitivity, and specificity of lists of differentially expressed genes in microarray studies (2008, background): “Criticism of and concerns about microarrays continue to appear in some of the most prestigious scientific journals [6-10], leading to a growing negative perception regarding microarray reproducibility, and hence reliability.”
The Influence of Feature Selection Methods on Accuracy, Stability and Interpretability of Molecular Signatures (2011, background): “Independently of differences in cohorts or technologies, [3] and [4] demonstrate that a major cause for the lack of overlap between signatures is that many different signatures lead to similar predictive accuracies, and that the process of estimating a signature is very sensitive to the samples used”
Subclass Mapping: Identifying Common Subtypes in Independent Disease Data Sets (2007, background): “DNA microarray-based whole genome expression profiling is subject to poor reproducibility of discovered molecular disease subtypes and can lead to biomarkers that do not generalize [1].”
Machine Learning and Its Applications to Biology (2007, background): “Because of inadequate validation schemes, many studies published in the literature as successful have been shown to be overoptimistic [40].”
The cell cycle and cancer (2012, background): “Disappointingly, the actual performance of prediction rules using gene expression has not been as informative as hoped for many tumour types and the list of genes identified can be highly unstable [22,23].”
A Survey on Filter Techniques for Feature Selection in Gene Expression Microarray Analysis (2012, background): “Several studies conducted for DEGs discovery show that many FS methods are highly dependent on the training set of samples [62], [63] resulting in lists of features which are unstable under variations in the training population.”