L1 penalized continuation ratio models for ordinal response prediction using high-dimensional datasets

Health status and outcomes are frequently measured on an ordinal scale. For high-throughput genomic datasets, the common approach to analyzing ordinal response data has been to break the problem into one or more dichotomous response analyses. This dichotomous response approach does not make use of all available data and therefore leads to loss of power and increases the number of type I errors. Herein we describe an innovative frequentist approach that combines two statistical techniques, L(1) penalization and continuation ratio models, for modeling an ordinal response using gene expression microarray data. We conducted a simulation study to assess the performance of two computational approaches and two model selection criteria for fitting frequentist L(1) penalized continuation ratio models. Moreover, we empirically compared the approaches using three application datasets, each of which seeks to classify an ordinal class using microarray gene expression data as the predictor variables. We conclude that the L(1) penalized constrained continuation ratio model is a useful approach for modeling an ordinal response for datasets where the number of covariates (p) exceeds the sample size (n) and the decision of whether to use Akaike Information Criterion (AIC) or Bayesian Information Criterion (BIC) for selecting the final model should depend upon the similarities between the pathologies underlying the disease states to be classified.     

Time-varying feature selection for longitudinal analysis

We propose time-varying coefficient model selection and estimation based on the spline approach, which is capable of capturing time-dependent covariate effects. The new penalty function utilizes local-region information for varying-coefficient estimation, in contrast to the traditional model selection approach focusing on the entire region. The proposed method is extremely useful when the signals associated with relevant predictors are time-dependent, and detecting relevant covariate effects in the local region is more scientifically relevant than those of the entire region. Our simulation studies indicate that the proposed model selection incorporating local features outperforms the global feature model selection approaches. The proposed method is also illustrated through a longitudinal growth and health study from National Heart, Lung, and Blood Institute.     

Identification of gene–environment interactions with marginal penalization

Gene–environment (G–E) interaction analysis has been extensively conducted for complex diseases. In marginal analysis, the common practice is to conduct likelihood-based (and other “standard”) estimation with each marginal model, and then select significant G–E interactions and main effects based on p values and multiple comparisons adjustment. One limitation of this approach is that the identification results often do not respect the “main effects, interactions” hierarchy, which has been stressed in recent G–E interaction analyses. There is some recent effort tackling this problem, however, with very complex formulations. Another limitation of the common practice is that it may not perform well when regularization is needed, for example, because of “non-normal” distributions. In this article, we propose a marginal penalization approach which adopts a novel penalty to directly tackle the aforementioned problems. The proposed approach has a framework more coherent with that of the recently developed joint analysis methods and an intuitive formulation, and can be effectively realized. In simulation, it outperforms the popular significance-based analysis and simple penalization-based alternatives. Promising findings are made in the analysis of a single-nucleotide polymorphism and a gene expression data.     

Analyzing Association Mapping in Pedigree‐Based GWAS Using a Penalized Multitrait Mixed Model

Genome‐wide association studies (GWAS) have led to the identification of many genetic variants associated with complex diseases in the past 10 years. Penalization methods, with significant numerical and statistical advantages, have been extensively adopted in analyzing GWAS. This study has been partly motivated by the analysis of Genetic Analysis Workshop (GAW) 18 data, which have two notable characteristics. First, the subjects are from a small number of pedigrees and hence related. Second, for each subject, multiple correlated traits have been measured. Most of the existing penalization methods assume independence between subjects and traits and can be suboptimal. There are a few methods in the literature based on mixed modeling that can accommodate correlations. However, they cannot fully accommodate the two types of correlations while conducting effective marker selection. In this study, we develop a penalized multitrait mixed modeling approach. It accommodates the two different types of correlations and includes several existing methods as special cases. Effective penalization is adopted for marker selection. Simulation demonstrates its satisfactory performance. The GAW 18 data are analyzed using the proposed method.     

Comparison of the efficacies of patching and penalization therapies for the treatment of amblyopia patients

AIM: Tocompare the efficacies of patching and penalization therapies for the treatment of amblyopia patients.METHODS:The records of 64 eyes of 50 patients 7 to 16y of age who had presented to our clinics with a diagnosis of amblyopia, were evaluated retrospectively. Forty eyes of 26 patients who had received patching therapy and 24 eyes of 24 patients who had received penalization therapy included in this study. The latencies and amplitudes of visual evoked potential (VEP) records and best corrected visual acuities (BCVA) of these two groups were compared before and six months after the treatment.RESULTS:In both patching and the penalization groups, the visual acuities increased significantly following the treatments (P<0.05). The latency measurements of the P100 wave obtained at 1.0°, 15 arc min. Patterns of both groups significantly decreased following the 6-months-treatment. However, the amplitude measurements increased (P<0.05).CONCLUSION: The patching and the penalization methods, which are the main methods used in the treatment of amblyopia, were also effective over the age of 7y, which has been accepted as the critical age for the treatment of amblyopia.     

A novel active contour model for serial image segmentation

A novel method for the segmentation of serial images is proposed. In the presented framework, the driving force acts as the attracting term to propel the evolving curve towards the object boundaries, and the adaptive term changes the sign of driving force accordingly. Therefore, the evolving curves can arrive at the desired direction without a requirement for the initial curve to be strictly inside or outside the object. A weighted length term is used to keep the smoothness of curve and penalize the formulation of discontinuities. To prevent the level set function deviating from a signed distance function, a distance rectifying flow is also added to the model; therefore the time-consuming re-initialization procedure is completely avoided. Experiments on both synthetic image and CT serial images demonstrate the feasibility and efficiency of the method.     

Gene–environment interaction analysis via deep learning

Gene–environment (G–E) interaction analysis plays an important role in studying complex diseases. Extensive methodological research has been conducted on G–E interaction analysis, and the existing methods are mostly based on regression techniques. In many fields including biomedicine and omics, it has been increasingly recognized that deep learning may outperform regression with its unique flexibility (e.g., in accommodating unspecified nonlinear effects) and superior prediction performance. However, there has been a lack of development in deep learning for G–E interaction analysis. In this article, we fill this important knowledge gap and develop a new analysis approach based on deep neural network in conjunction with penalization. The proposed approach can simultaneously conduct model estimation and selection (of important main G effects and G–E interactions), while uniquely respecting the “main effects, interactions” variable selection hierarchy. Simulation shows that it has superior prediction and feature selection performance. The analysis of data on lung adenocarcinoma and skin cutaneous melanoma overall survival further establishes its practical utility. Overall, this study can advance G–E interaction analysis by delivering a powerful new analysis approach based on modern deep learning.     

大龄儿童及青少年弱视治疗的观察

目的探讨大龄儿童及青少年弱视治疗效果。方法观察9~16岁弱视77例85只眼,采用光学药物压抑疗法加遮盖疗法联合红光闪烁等综合治疗,平均随访26个月。结果 85只弱视眼中基本治愈52.94%(45只眼),进步32.94%(28只眼),无效14.12%(12只眼);按年龄分组,9-12岁组53只眼,基本治愈31只眼(58.49%),13～16岁组32只眼,基本14只眼(43.75%),两组差异无统计学意义(P>0.05)。对13~16岁组进一步分析,其疗效与注视性质、弱视类型、弱视程度有关。结论大龄儿童及青少年弱视大部分治疗是有效的。     

Integrative sparse principal component analysis of gene expression data

In the analysis of gene expression data, dimension reduction techniques have been extensively adopted. The most popular one is perhaps the PCA (principal component analysis). To generate more reliable and more interpretable results, the SPCA (sparse PCA) technique has been developed. With the “small sample size, high dimensionality” characteristic of gene expression data, the analysis results generated from a single dataset are often unsatisfactory. Under contexts other than dimension reduction, integrative analysis techniques, which jointly analyze the raw data of multiple independent datasets, have been developed and shown to outperform “classic” meta‐analysis and other multidatasets techniques and single‐dataset analysis. In this study, we conduct integrative analysis by developing the iSPCA (integrative SPCA) method. iSPCA achieves the selection and estimation of sparse loadings using a group penalty. To take advantage of the similarity across datasets and generate more accurate results, we further impose contrasted penalties. Different penalties are proposed to accommodate different data conditions. Extensive simulations show that iSPCA outperforms the alternatives under a wide spectrum of settings. The analysis of breast cancer and pancreatic cancer data further shows iSPCA's satisfactory performance.