Semi-Supervised Methods to Predict Patient Survival from Gene Expression Data

By Eric Bair, Robert Tibshirani. PLoS Biology, 2004

756 citations recorded by the source corpus.

An important goal of DNA microarray research is to develop tools to diagnose cancer more accurately based on the genetic profile of a tumor. There are several existing techniques in the literature for performing this type of diagnosis. Unfortunately, most of these techniques assume that different subtypes of cancer are already known to exist. Their utility is limited when such subtypes have not been previously identified. Although methods for identifying such subtypes exist, these methods do not work well for all datasets. It would be desirable to develop a procedure to find such subtypes that is applicable in a wide variety of circumstances. Even if no information is known about possible subtypes of a certain form of cancer, clinical information about the patients, such as their survival time, is often available. In this study, we develop some procedures that utilize both the gene expression data and the clinical data to identify subtypes of cancer and use this knowledge to diagnose future patients. These procedures were successfully applied to several publicly available datasets. We present diagnostic procedures that accurately predict the survival of future patients based on the gene expression profile and survival times of previous patients. This has the potential to be a powerful tool for diagnosing and treating cancer.

Subfields: Cancer Research, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/6cb2ebb02d506dbcb0e004037d72e4746cf573da

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

Quantile-adaptive model-free variable screening for high-dimensional heterogeneous data (2013, methodology): “To assess the predictive performance of the proposed method, we divide the data set into a training set with n1 = 160 patients and a testing set with remaining n2 = 80 patients, in the same way as Bair and Tibshirani (2004) did. The index of the training set is available from http://www-stat.stanfor”
A Review of Feature Selection and Feature Extraction Methods Applied on Microarray Data (2015, background): “It has been argued [65, 66] that when computing the principal components (PCs) of a dataset there is no guarantee that the PCs will be related to the class variable.”
Penalized Cox regression analysis in the high-dimensional and low-sample size settings, with applications to microarray gene expression data (2005, methodology): “A comparison with other methods As a comparison, we also analyzed the lymphoma dataset using three other methods, the PCR method of Li and Gui (2004), the L2 penalized method of Li and Luan (2003) and the SPCA method of Bair and Tibshirani (2004) .”
DeepSurv: personalized treatment recommender system using a Cox proportional hazards deep neural network (2016, background): “It is a common practice in medical applications to determine the relationship between a patient’s observable covariates and his or her risk of an event [13, 3, 26].”
Survival ensembles. (2006, neutral): “In contrast, dimension reduction techniques are studied by Li and Li (2004) and Bair and Tibshirani (2004), who advocate the application of low-dimensional compound covariates obtained from an unsupervised clustering of the covariates.”
Systemic Immunity is Required for Effective Cancer Immunotherapy (2017, neutral): “We therefore determined whether the statistical inference integrated into Citrus could, instead, be applied to Scaffold maps.”
Deep Survival : The Deep Cox Proportional Hazards Network (2016, background): “This poses a problem in medical applications, as determining the significance of patient’s factors on their risk is common practice in survival analysis [9, 3, 21].”
Analysis of Gene Expression Data Using BRB-Array Tools (2007, methodology): “This tool is based on proportional hazards modeling in which the expression data is represented by the fi rst several principal components of the genes whose expression is correlated with survival outcome, i.e. supervised principal components (Bair and Tibshirani, 2004).”
Whole slide images based cancer survival prediction using attention guided deep multiple instance learning networks (2020, background): “5 years)? What rate will they die or fail? It is a very important clinical application and many efforts have been made to search for biomarkers from omics data that are significantly related to patient death [1–5].”
GSVA: gene set variation analysis for microarray and RNA-Seq data (2013, methodology): “In a similar way to the survival simulation by Bair and Tibshirani [36], we generated survival times and censoring status for each observation with different parameters for each group of samples (see Methods).”
The International Classification of Headache Disorders, 3rd edition (beta version) (2013, methodology): “Finally, an Axis IV is envisioned as a method to classify a patient into clinically meaningful categories by collapsing large amounts of variability across biopsychosocial and molecular genomic domains through, for example, the use of modern clustering models.(128) Although the DC/TMD will be in imp”
An extensive experimental survey of regression methods (2019, methodology): “This method [45] 795 retains only a subset of the principal components which are correlated 796 to the output.”
A computational solution for bolstering reliability of epigenetic clocks: Implications for clinical trials and longitudinal tracking (2021, neutral): “12) 68,69 .”
High-Dimensional Variable Selection for Survival Data (2010, methodology): “Some methods implicitly use Cox regression by making use of Cox scores; for example, “corrected” Cox scores were used by Bair and Tibshirani (2004) for semisupervised prediction using principal component regression and by Tibshirani et al. (2002) for semisupervised classification using…”
Nearest Template Prediction: A Single-Sample-Based Flexible Class Prediction with Confidence Assessment (2010, methodology): “, these methods merely split the dataset into equal-sized subsets using a cut-off such as median specific to the sample collection, therefore each prediction could be changed according to which samples are included in the collection [7,23,24].”
Independent screening for single‐index hazard rate models with ultrahigh dimensional features (2011, methodology): “Based the method of supervised principal components (Bair and Tibshirani, 2004), the gene expressions were used to develop an 86-gene signature for predicting survival.”