Dataset Information

Prediction of survival with alternative modeling techniques using pseudo values.

ABSTRACT: BACKGROUND:The use of alternative modeling techniques for predicting patient survival is complicated by the fact that some alternative techniques cannot readily deal with censoring, which is essential for analyzing survival data. In the current study, we aimed to demonstrate that pseudo values enable statistically appropriate analyses of survival outcomes when used in seven alternative modeling techniques. METHODS:In this case study, we analyzed survival of 1282 Dutch patients with newly diagnosed Head and Neck Squamous Cell Carcinoma (HNSCC) with conventional Kaplan-Meier and Cox regression analysis. We subsequently calculated pseudo values to reflect the individual survival patterns. We used these pseudo values to compare recursive partitioning (RPART), neural nets (NNET), logistic regression (LR) general linear models (GLM) and three variants of support vector machines (SVM) with respect to dichotomous 60-month survival, and continuous pseudo values at 60 months or estimated survival time. We used the area under the ROC curve (AUC) and the root of the mean squared error (RMSE) to compare the performance of these models using bootstrap validation. RESULTS:Of a total of 1282 patients, 986 patients died during a median follow-up of 66 months (60-month survival: 52% [95% CI: 50%-55%]). The LR model had the highest optimism corrected AUC (0.791) to predict 60-month survival, followed by the SVM model with a linear kernel (AUC 0.787). The GLM model had the smallest optimism corrected RMSE when continuous pseudo values were considered for 60-month survival or the estimated survival time followed by SVM models with a linear kernel. The estimated importance of predictors varied substantially by the specific aspect of survival studied and modeling technique used. CONCLUSIONS:The use of pseudo values makes it readily possible to apply alternative modeling techniques to survival problems, to compare their performance and to search further for promising alternative modeling techniques to analyze survival time.

SUBMITTER: van der Ploeg T

PROVIDER: S-EPMC4065009 | biostudies-literature | 2014

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Prediction of survival with alternative modeling techniques using pseudo values.

van der Ploeg Tjeerd T Datema Frank F Baatenburg de Jong Robert R Steyerberg Ewout W EW

PloS one 20140620 6

<h4>Background</h4>The use of alternative modeling techniques for predicting patient survival is complicated by the fact that some alternative techniques cannot readily deal with censoring, which is essential for analyzing survival data. In the current study, we aimed to demonstrate that pseudo values enable statistically appropriate analyses of survival outcomes when used in seven alternative modeling techniques.<h4>Methods</h4>In this case study, we analyzed survival of 1282 Dutch patients wit ...[more]

PMID: 24950066

Similar Datasets

Project description:Recently it was shown that gene expression signatures generated from DNA microarray analyses have promise as biomarkers of clinical outcome and that the molecular characteristics of tumors could be elucidated. Through this study, we have determined a high-risk signature for recurrence as a prognostic biomarker using formalin-fixed HNSCC tumors and tested the results to an independent data set obtained from fresh frozen tumors as a comparison. Also, we have shown the genes that are involved in epithelial to mesenchymal transition and nuclear factor-ĸB signaling deregulation are the most prominent molecular characteristics of the high-risk tumors. Forty samples including 34 formalin-fixed tissues and 6 matched frozen tissues from 29 HNSCC patients were analyzed for gene expression. The formalin-fixed tumors were classified based on their gene expression by intrinsic analysis and the intrinsic gene list was tested on the classification of previously published 60 frozen HNSCC tumors which highly correlated. Based on the molecular classification, a 75-gene list that is predictive of high-risk for recurrence was determined by training on the formalin-fixed tumor set and tested on the independent frozen tumor set. The difference in recurrence-free survival (RFS) between the high-risk vs. low-risk groups in the training and test sets were statistically significant (Log-rank test, p=0.002 and p=0.03, respectively). Also, the gene expression data was interrogated using Gene Set Enrichment Analysis to determine functional significance. The most significant sets of genes that are enriched in the high-risk tumors were genes involving epithelial to mesenchymal transition (EMT), NF-ĸB activation and cell adhesion. In conclusion, global gene expression analysis is feasible using formalin-fixed tissue, and the data from different sample preparations and array platforms can be reliably combined for analyses. The 75-gene list can be utilized as a prognostic biomarker of recurrence and the molecular characteristics of EMT and NF-ĸB activation can be targeted as the novel therapy in the identified high-risk patients. Keywords: FFPE, survival analysis

Project description:BACKGROUND:Cyclic peptide-based drug discovery is attracting increasing interest owing to its potential to avoid target protein depletion. In drug discovery, it is important to maintain the biostability of a drug within the proper range. Plasma protein binding (PPB) is the most important index of biostability, and developing a computational method to predict PPB of drug candidate compounds contributes to the acceleration of drug discovery research. PPB prediction of small molecule drug compounds using machine learning has been conducted thus far; however, no study has investigated cyclic peptides because experimental information of cyclic peptides is scarce. RESULTS:First, we adopted sparse modeling and small molecule information to construct a PPB prediction model for cyclic peptides. As cyclic peptide data are limited, applying multidimensional nonlinear models involves concerns regarding overfitting. However, models constructed by sparse modeling can avoid overfitting, offering high generalization performance and interpretability. More than 1000 PPB data of small molecules are available, and we used them to construct a prediction models with two enumeration methods: enumerating lasso solutions (ELS) and forward beam search (FBS). The accuracies of the prediction models constructed by ELS and FBS were equal to or better than those of conventional non-linear models (MAE?=?0.167-0.174) on cross-validation of a small molecule compound dataset. Moreover, we showed that the prediction accuracies for cyclic peptides were close to those for small molecule compounds (MAE?=?0.194-0.288). Such high accuracy could not be obtained by a simple method of learning from cyclic peptide data directly by lasso regression (MAE?=?0.286-0.671) or ridge regression (MAE?=?0.244-0.354). CONCLUSION:In this study, we proposed a machine learning techniques that uses low-dimensional sparse modeling to predict the PPB value of cyclic peptides computationally. The low-dimensional sparse model not only exhibits excellent generalization performance but also improves interpretation of the prediction model. This can provide common an noteworthy knowledge for future cyclic peptide drug discovery studies.

Project description:Potato breeding must improve its efficiency by increasing the reliability of selection as well as identifying a promising germplasm for crossing. This study shows the prediction accuracy of genomic-estimated breeding values for several potato (Solanum tuberosum L.) breeding clones and the released cultivars that were evaluated at three locations in northern and southern Sweden for various traits. Three dosages of marker alleles [pseudo-diploid (A), additive tetrasomic polyploidy (B), and additive-non-additive tetrasomic polyploidy (C)] were considered in the genome-based prediction models, for single environments and multiple environments (accounting for the genotype-by-environment interaction or G × E), and for comparing two kernels, the conventional linear, Genomic Best Linear Unbiased Prediction (GBLUP) (GB), and the non-linear Gaussian kernel (GK), when used with the single-kernel genetic matrices of A, B, C, or when employing two-kernel genetic matrices in the model using the kernels from B and C for a single environment (models 1 and 2, respectively), and for multi-environments (models 3 and 4, respectively). Concerning the single site analyses, the trait with the highest prediction accuracy for all sites under A, B, C for model 1, model 2, and for GB and GK methods was tuber starch percentage. Another trait with relatively high prediction accuracy was the total tuber weight. Results show an increase in prediction accuracy of model 2 over model 1. Non-linear Gaussian kernel (GK) did not show any clear advantage over the linear kernel GBLUP (GB). Results from the multi-environments had prediction accuracy estimates (models 3 and 4) higher than those obtained from the single-environment analyses. Model 4 with GB was the best method in combination with the marker structure B for predicting most of the tuber traits. Most of the traits gave relatively high prediction accuracy under this combination of marker structure (A, B, C, and B-C), and methods GB and GK combined with the multi-environment with G × E model.

Dataset Information

Prediction of survival with alternative modeling techniques using pseudo values.

Publications

Prediction of survival with alternative modeling techniques using pseudo values.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets