Cawley GC, Talbot NL, et al (2006) (14)
|
foodborne microbial pathogen
|
Log-rank test, Cox proportional hazards models, Cox-nnet, Kaplan-Meier estimator.
|
quadratic or cubic
polynomial, Gaussian, radial basis function (RBF) kernel, Bayesian kernel-accelerated life survival analysis
|
Introducing a parametric accelerated life survival analysis model based on kernel learning methods that, at least in principle, can learn arbitrary dependencies between a vector of explanatory variables and the scale of the distribution of survival times.
|
The kernel survival analysis models are found to be more accurate than models based on more traditional survival analysis techniques but also suggested a risk assessment of the foodborne botulism hazard would benefit from the collection of additional data.
|
Srivastava S, Wang WY et al (2013) (9)
|
Glioblastoma Multiform (GBM)
|
Kaplan-Meier estimates
|
Gaussian radial basis function (RBF) kernel
|
Predicting relevant clinical outcomes, proposed a flexible statistical machine learning approach that acknowledges and models the interaction between platform-specific measurements through nonlinear kernel machines and borrows information within and between platforms through a hierarchical Bayesian framework.
|
In terms of prediction accuracy, it showed that non-linear and interaction-based integrative methods perform better than linear alternatives and non-integrative methods that do not account for interactions between the platforms.
|
Seoane JA et al (2013)
(56)
|
breast cancer
|
Cox regression
|
Composite kernels, linear kernels, polynomial base kernels, and non-linear Gaussian kernels.
|
Considering a variety of learning strategies to boost prediction performance based on the use of all available data.
|
Predictive accuracy is improved by using kernels that exclusively use those genes which are known members of particular pathways.
|
Speicher NK, Pfeifer N (2015)
(41)
|
five different cancer sets (glioblastoma multiforme, breast invasive carcinoma, kidney renal clear cell carcinoma, lung squamous cell carcinoma, colon adenocarcinoma)
|
the Cox regression model
|
The Gaussian radial basis kernel function
|
Integrating different data types by regularized unsupervised multiple kernels learning with application to cancer subtype discovery
|
Input kernel matrices with only a little information have less impact on the integrated kernel matrix. The visualizations of the contributions of the individual kernels suggest that using more than one kernel matrix per data type can even be beneficial, and the stability analysis shows that the method does not overfit when more kernels are added.
|
Kiaee F et al (2016)
(34)
|
pancreas-head cancer, lung cancer, prostatic cancer, breast
cancer
|
Cox PH model, Kaplan–Meier curves, Accelerated failure time (AFT) model
|
non-Mercer kernels, Bayesian kernels, Gaussian kernels, regression-based
kernel survival models
|
Flexibility to utilize arbitrary numbers and types of kernel functions (e.g., non-Mercer kernels and multi-kernel learning).
|
1) extra sparsity that leads to a better generalization and avoids overfitting; 2) automatic relevance sample determination based on data that provide more accuracy, in particular for highly censored survival data; and 3) flexibility to utilize arbitrary numbers and types of kernel functions.
|
Speicher NK, Pfeifer N (2017)
(60)
|
Cancer subtyping (breast, colon, glioblastoma, kidney, lung)
|
The log-rank test of the Cox regression model.
|
the Gaussian radial basis kernel function,
|
Proposing an unsupervised data integration method based on kernel principal component analysis.
|
It proposed a scoring function to determine the best combination of the input data. On five cancer data sets, it showed that this procedure works in most cases better or as well as naive approaches in terms of survival analysis.
|
Sinnott JA, Cai T (2018)
(43)
|
ovarian cancer, Breast Cancer
|
the Cox
proportional hazards and semiparametric accelerated failure time models
|
Gaussian kernel.
|
proposing MKL methods for censored survival outcomes
|
The proposed MKL-based prediction methods work well in the finite sample and can potentially outperform models constructed assuming linear effects or ignoring the group knowledge.
|
Sun D, Li A et al (2018)
(44)
|
Breast cancer
|
The Cox proportion hazards model, LASSO-Cox, elastic-net penalized Cox (EN-Cox)
models
|
geneExpr kernel, CAN kernel, methyl kernel, and protein-kernel
|
Introducing a novel method called GPMKL based on multiple kernel learning (MKL), which efficiently employs heterogeneous information containing genomic data and pathological images.
|
The proposed framework with pathological images performs remarkably better than the existing survival prediction methods.
|
Onur Dereli (2019)
(45)
|
the Genomics Data
Commons (GDC) data
|
Survival Random forest, Survival SVM algorithm
|
linear, polynomial, and Gaussian
kernels
|
Picking the predictive pathways/gene sets from a given collection and then training a predictive model on the subset of genomic features mapped to these selected pathways/gene sets and developing a novel machine learning algorithm (Path2Surv) that conjointly performs these two steps using multiple kernel learning.
|
Path2Surv statistically significantly outperformed survival random forest (RF) on 12 out of 20 datasets and obtained comparable predictive performance against survival support vector machine (SVM) using significantly fewer gene expression features (i.e. less than 10% of what survival RF and survival SVM used).
|
Zhang A (2019)
(46)
|
lung cancer
|
Cox proportional hazards model, Parametric
censored regression models (PCRM), Elastic-net penalized Cox
(EN-Cox), LASSO-Cox, Random survival forests (RSF)
|
Gaussian kernel, linear kernel
|
Proposing a novel method called LSCDFS-MKL, which is based on multiple kernel learning to predict DFS of lung SCC.
|
The features extracted from pathological images play an important role in DFS prediction of lung SCC and
LSCDFS-MKL has a significantly better performance than other prediction methods.
|
Pan Z, Zhang H, et al (2019)
(54)
|
Head and neck squamous cell carcinoma (HNSC)
|
Kaplan-Meier survival analysis, Wilcoxon signed-rank test
|
Gaussian kernels
|
Proposing a novel self-weighted, multi-kernel, multi-label learning (SwMKML) method to predict disease-related miRNAs.
|
Compared with four state-of-the-art computational models, SwMKML achieved the best results of 95.5%, 93.1%, and 84.1% in global leave-one-out cross-validation, 5-fold cross-validation, and overall prediction accuracy, respectively.
|
Gong C, Hu Y et al (2019)
(55)
|
gastric cancer
|
Kaplan–Meier estimate
|
Multiple kernels learning-based Kronecker regularized least squares.
|
Identifying potential vital genes and modules involved in the progression of GC.
|
The gene expression profile of the GCPR module was mapped connectivity map (Cmap), and the mTOR inhibitor (Sirolimus) was found to be the most promising candidate.
|
Röder B, Kersten N et al (2019)
(58)
|
head and neck squamous cell carcinoma
|
the Kaplan–Meier estimation, log-rank test,
|
RBF, polynomial, and linear kernel
|
Presenting web-rMKL, a web server that provides an integrative dimensionality reduction with subsequent clustering of samples based on data from multiple inputs.
|
Introduced a preprocessing tool for generating kernel matrices from tables containing numerical feature values. This program can be used to generate admissible input if no precomputed kernel matrices are available.
|
Zhang Y, Li A et al (2019)
(61)
|
brain tumors
|
Kaplan-Meier estimate plot, Cox model,
|
the Gaussian RBF kernel
|
Developing a novel MKL method, named HI-MKL that could integrate both histopathological images and multi-omics data efficiently.
|
HI-MKL is an accurate, robust, and generalized MKL method that performs well in GBM prognosis tasks.
|
He Z, Zhang J et al (2020)
(42)
|
Breast cancer
|
the Cox regression model
|
polynomial base kernel, Gaussian kernel
|
Adopting multiple kernel learning (MKL) to efficiently integrate somatic mutation to current molecular data including gene expression, copy number variation (CNV), methylation, and protein expression data for the prediction of breast cancer survival.
|
MKL outperformed the other traditional classifiers in multi-omics data integration. Through employing promising omics data such as somatic mutations and harnessing the power of proper feature selection methods and effective integration frameworks, breast cancer survival predictive accuracy can be further increased.
|
Molstad AJ et al (2021)
(35)
|
Kidney renal clear cell carcinoma
|
Cox proportional
hazards models. accelerated failure time model (AFT)
|
radial basis function kernel, genome-wide kernel, pathway kernels
|
Proposing a Gaussian process accelerated failure time model to predict survival time using genome-wide or pathway-wide gene expression data.
|
Defining multiple kernels using gene expression from multiple pathways, though, in practice, these kernels can be defined using the same set of genes with different definitions of distance/similarities, or they can be defined based on multiple types of -omic data. Although this has developed its method for survival time prediction, it can be used or extended for other outcomes with certain patterns of missing data.
|
Feng J, Jiang L et al (2021)
(31)
|
seven cancer datasets (liver, colon, stomach, breast, lung, kidney, Blood)
|
Cox regression model
|
polynomial kernel, Gaussian kernel, fused kernel, kernel principal component analysis (PCA)
|
Taking kernel PCA to extract features for each profile, then converting them into three similarity kernel matrices, and fusing them into one. Finally, apply it to the spectral clustering algorithm to get the clustering results of different cancer subtypes.
|
Using kernel PCA for feature reduction, taking weighted fusion strategies instead of complex SKF, and finally generating clustering results from spectral clustering had comparable reliability and stability.
|
Wilson CM, Li K et al (2021)
(10)
|
Skin Cutaneous Melanoma
(SKCM) gene and miRNA expression data
|
Cox proportional hazard model, accelerated failure time (AFT).
|
Linear, Polynomial, radial base function (RBF) kernel
|
Showing that a convex conjugate function of the Cox loss function based on Fenchel duality exists, and provides an alternative framework to optimization based on the primal form.
|
An efficient algorithm for solving the kernel learning problem with censored survival outcomes.
|
Park S et al (2021)
(48)
|
Breast cancer, Pancreatic cancer
|
Log-rank test
|
Gaussian kernel
|
Proposing a kernel-learning method to iteratively update the importance of each view (data set) and performing refined clustering analysis by aggregating multiple omics data sets that include a large number of molecular features.
|
The statistical properties of the proposed multi-view clustering method showed that it can accurately infer the underlying clusters under some regularity conditions.
|
Uzunangelov V et al (2021)
(59)
|
Colorectal cancer, breast cancer.
|
paired Wilcoxon signed rank test, Cox regression,
|
RF kernel, Gaussian kernel, polynomial kernels
|
Introducing the Algorithm for Kernel Learning with Integrative Modules of Approximating Tree Ensembles (AKLIMATE)—a novel approach that combines heterogeneous data with prior knowledge in the form of gene sets.
|
Demonstrated the method performs as well as, or better than, top approaches on three very different cancer genomics applications.
|