Search

Learning predictive models in new domains with scarce training data is a growing challenge in modern supervised learning scenarios. This incentivizes developing domain adaptation methods that leverage the knowledge in known domains (source) and adapt to new domains (target) with a different probability distribution. This becomes more challenging when the source and target domains are in heterogeneous feature spaces, known as heterogeneous domain adaptation (HDA). While most HDA methods utilize mathematical optimization to map source and target data to a common space, they suffer from low transferability. Neural representations have proven to be more transferable; however, they are mainly designed for homogeneous environments. Drawing on the theory of domain adaptation, we propose a novel framework, Heterogeneous Adversarial Neural Domain Adaptation (HANDA), to effectively maximize the transferability in heterogeneous environments. HANDA conducts feature and distribution alignment in a unified neural network architecture and achieves domain invariance through adversarial kernel learning. Three experiments were conducted to evaluate the performance against the state-of-the-art HDA methods on major image and text e-commerce benchmarks. HANDA shows statistically significant improvement in predictive performance. The practical utility of HANDA was shown in real-world dark web online markets. HANDA is an important step towards successful domain adaptation in e-commerce applications., Comment: Forthcoming in IEEE Transactions on Pattern Recognition and Machine Intelligence (TPAMI)

Spatial regression is widely used for modeling the relationship between a dependent variable and explanatory covariates. Oftentimes, the linear relationships vary across space, when some covariates have location-specific effects on the response. One fundamental question is how to detect the systematic variation in the model and identify which locations share common regression coefficients and which do not. Only a correct model structure can assure unbiased estimation of coefficients and valid inferences. In this work, we propose a new procedure, called Spatial Heterogeneity Automatic Detection and Estimation (SHADE), for automatically and simultaneously subgrouping and estimating covariate effects for spatial regression models. The SHADE employs a class of spatially-weighted fusion type penalty on all pairs of observations, with location-specific weight adaptively constructed using spatial information, to cluster coefficients into subgroups. Under certain regularity conditions, the SHADE is shown to be able to identify the true model structure with probability approaching one and estimate regression coefficients consistently. We develop an alternating direction method of multiplier algorithm (ADMM) to compute the SHAD efficiently. In numerical studies, we demonstrate empirical performance of the SHADE by using different choices of weights and compare their accuracy. The results suggest that spatial information can enhance subgroup structure analysis in challenging situations when the spatial variation among regression coefficients is small or the number of repeated measures is small. Finally, the SHADE is applied to find the relationship between a natural resource survey and a land cover data layer to identify spatially interpretable groups.

Multiclass probability estimation is the problem of estimating conditional probabilities of a data point belonging to a class given its covariate information. It has broad applications in statistical analysis and data science. Recently a class of weighted Support Vector Machines (wSVMs) has been developed to estimate class probabilities through ensemble learning for K-class problems (Wu et al., 2010; Wang et al., 2019), where K is the number of classes. The estimators are robust and achieve high accuracy for probability estimation, but their learning is implemented through pairwise coupling, which demands polynomial time in K. In this paper, we propose two new learning schemes, the baseline learning and the One-vs-All (OVA) learning, to further improve wSVMs in terms of computational efficiency and estimation accuracy. In particular, the baseline learning has optimal computational complexity in the sense that it is linear in K. Though not the most efficient in computation, the OVA is found to have the best estimation accuracy among all the procedures under comparison. The resulting estimators are distribution-free and shown to be consistent. We further conduct extensive numerical experiments to demonstrate their finite sample performance.

In this communication, we propose using modern machine learning (ML) techniques including least absolute shrinkage and selection operator (lasso), artificial neural networks (ANNs), and $k$ -nearest neighbor (kNN) methods for antenna design optimization. The automated techniques are shown to provide an efficient, flexible, and reliable framework to identify optimal design parameters for a reference dual-band double T-shaped monopole antenna to achieve favorite performance in terms of its two bands, i.e., between 2.4 and 3.0 and 5.15 and 5.6 GHz. In this communication, we also present a thorough study and comparative analysis of the results predicted by these ML techniques, with the results obtained from high-frequency structure simulator (HFSS) to verify the accuracy of these techniques.

There is an urgent need to identify the cellular and molecular mechanisms responsible for severe COVID-19 that results in death. We initially performed both untargeted and targeted lipidomics as well as focused biochemical analyses of 127 plasma samples and found elevated metabolites associated with secreted phospholipase A(2) (sPLA(2)) activity and mitochondrial dysfunction in patients with severe COVID-19. Deceased COVID-19 patients had higher levels of circulating, catalytically active sPLA(2) group IIA (sPLA(2)-IIA), with a median value that was 9.6-fold higher than that for patients with mild disease and 5.0-fold higher than the median value for survivors of severe COVID-19. Elevated sPLA(2)-IIA levels paralleled several indices of COVID-19 disease severity (e.g., kidney dysfunction, hypoxia, multiple organ dysfunction). A decision tree generated by machine learning identified sPLA(2)-IIA levels as a central node in the stratification of patients who died from COVID-19. Random forest analysis and least absolute shrinkage and selection operator–based (LASSO-based) regression analysis additionally identified sPLA(2)-IIA and blood urea nitrogen (BUN) as the key variables among 80 clinical indices in predicting COVID-19 mortality. The combined PLA-BUN index performed significantly better than did either one alone. An independent cohort (n = 154) confirmed higher plasma sPLA(2)-IIA levels in deceased patients compared with levels in plasma from patients with severe or mild COVID-19, with the PLA-BUN index–based decision tree satisfactorily stratifying patients with mild, severe, or fatal COVID-19. With clinically tested inhibitors available, this study identifies sPLA(2)-IIA as a therapeutic target to reduce COVID-19 mortality.

Because different patients may respond quite differently to the same drug or treatment, there is an increasing interest in discovering individualized treatment rules. In particular, there is an emerging need to find optimal individualized treatment rules, which would lead to the "best" clinical outcome. In this paper, we propose a new class of loss functions and estimators based on robust regression to estimate the optimal individualized treatment rules. Compared to existing estimation methods in the literature, the new estimators are novel and advantageous in the following aspects. First, they are robust against skewed, heterogeneous, heavy-tailed errors or outliers in data. Second, they are robust against a misspecification of the baseline function. Third, under some general situations, the new estimator coupled with the pinball loss approximately maximizes the outcome's conditional quantile instead of the conditional mean, which leads to a more robust optimal individualized treatment rule than the traditional mean-based estimators. Consistency and asymptotic normality of the proposed estimators are established. Their empirical performance is demonstrated via extensive simulation studies and an analysis of an AIDS data set.

There is an urgent need to identify cellular and molecular mechanisms responsible for severe COVID-19 disease accompanied by multiple organ failure and high mortality rates. Here, we performed untargeted/targeted lipidomics and focused biochemistry on 127 patient plasma samples, and showed high levels of circulating, enzymatically active secreted phospholipase A2 Group IIA (sPLA2-IIA) in severe and fatal COVID-19 disease compared with uninfected patients or mild illness. Machine learning demonstrated that sPLA2-IIA effectively stratifies severe from fatal COVID-19 disease. We further introduce a PLA-BUN index that combines sPLA2-IIA and blood urea nitrogen (BUN) threshold levels as a critical risk factor for mitochondrial dysfunction, sustained inflammatory injury and lethal COVID-19. With the availability of clinically tested inhibitors of sPLA2-IIA, our study opens the door to a precision intervention using indices discovered here to reduce COVID-19 mortality.

In this era of data science-driven bioinformatics, machine learning research has focused on feature selection as users want more interpretation and post-hoc analyses for biomarker detection. However, when there are more features (i.e., transcripts) than samples (i.e., mice or human samples) in a study, it poses major statistical challenges in biomarker detection tasks as traditional statistical techniques are underpowered in high dimension. Second and third order interactions of these features pose a substantial combinatoric dimensional challenge. In computational biology, random forest (RF) classifiers are widely used due to their flexibility, powerful performance, their ability to rank features, and their robustness to the "P N" high-dimensional limitation that many matrix regression algorithms face. We propose binomialRF, a feature selection technique in RFs that provides an alternative interpretation for features using a correlated binomial distribution and scales efficiently to analyze multiway interactions.In both simulations and validation studies using datasets from the TCGA and UCI repositories, binomialRF showed computational gains (up to 5 to 300 times faster) while maintaining competitive variable precision and recall in identifying biomarkers' main effects and interactions. In two clinical studies, the binomialRF algorithm prioritizes previously-published relevant pathological molecular mechanisms (features) with high classification precision and recall using features alone, as well as with their statistical interactions alone.binomialRF extends upon previous methods for identifying interpretable features in RFs and brings them together under a correlated binomial distribution to create an efficient hypothesis testing algorithm that identifies biomarkers' main effects and interactions. Preliminary results in simulations demonstrate computational gains while retaining competitive model selection and classification accuracies. Future work will extend this framework to incorporate ontologies that provide pathway-level feature selection from gene expression input data.

ObjectiveTo develop and validate a machine-learning algorithm to improve prediction of incident OUD diagnosis among Medicare beneficiaries with ≥1 opioid prescriptions.MethodsThis prognostic study included 361,527 fee-for-service Medicare beneficiaries, without cancer, filling ≥1 opioid prescriptions from 2011-2016. We randomly divided beneficiaries into training, testing, and validation samples. We measured 269 potential predictors including socio-demographics, health status, patterns of opioid use, and provider-level and regional-level factors in 3-month periods, starting from three months before initiating opioids until development of OUD, loss of follow-up or end of 2016. The primary outcome was a recorded OUD diagnosis or initiating methadone or buprenorphine for OUD as proxy of incident OUD. We applied elastic net, random forests, gradient boosting machine, and deep neural network to predict OUD in the subsequent three months. We assessed prediction performance using C-statistics and other metrics (e.g., number needed to evaluate to identify an individual with OUD [NNE]). Beneficiaries were stratified into subgroups by risk-score decile.ResultsThe training (n = 120,474), testing (n = 120,556), and validation (n = 120,497) samples had similar characteristics (age ≥65 years = 81.1%; female = 61.3%; white = 83.5%; with disability eligibility = 25.5%; 1.5% had incident OUD). In the validation sample, the four approaches had similar prediction performances (C-statistic ranged from 0.874 to 0.882); elastic net required the fewest predictors (n = 48). Using the elastic net algorithm, individuals in the top decile of risk (15.8% [n = 19,047] of validation cohort) had a positive predictive value of 0.96%, negative predictive value of 99.7%, and NNE of 104. Nearly 70% of individuals with incident OUD were in the top two deciles (n = 37,078), having highest incident OUD (36 to 301 per 10,000 beneficiaries). Individuals in the bottom eight deciles (n = 83,419) had minimal incident OUD (3 to 28 per 10,000).ConclusionsMachine-learning algorithms improve risk prediction and risk stratification of incident OUD in Medicare beneficiaries.

The problem of interaction selection in high-dimensional data analysis has recently received much attention. This note aims to address and clarify several fundamental issues in interaction selection for linear regression models, especially when the input dimension p is much larger than the sample size n. We first discuss how to give a formal definition of “importance” for main and interaction effects. Then we focus on two-stage methods, which are computationally attractive for high-dimensional data analysis but thus far have been regarded as heuristic. We revisit the counterexample of Turlach and provide new insight to justify two-stage methods from the theoretical perspective. In the end, we suggest new strategies for interaction selection under the marginality principle and provide some simulation results.

Reactivating quiescent cells to proliferate is critical to tissue repair and homoeostasis. Quiescence exit is highly noisy even for genetically identical cells under the same environmental conditions. Deregulation of quiescence exit is associated with many diseases, but cellular mechanisms underlying the noisy process of exiting quiescence are poorly understood. Here we show that the heterogeneity of quiescence exit reflects a memory of preceding cell growth at quiescence induction and immediate division history before quiescence entry, and that such a memory is reflected in cell size at a coarse scale. The deterministic memory effects of preceding cell cycle, coupled with the stochastic dynamics of an Rb-E2F bistable switch, jointly and quantitatively explain quiescence-exit heterogeneity. As such, quiescence can be defined as a distinct state outside of the cell cycle while displaying a sequential cell order reflecting preceding cell growth and division variations., The quiescence-exit process is noisy even in genetically identical cells under the same environmental conditions. Here the authors show that the heterogeneity of quiescence exit reflects a memory of preceding cell growth at quiescence induction and immediate division history prior to quiescence entry.

The reactivation of quiescent cells to proliferate is fundamental to tissue repair and homeostasis in the body. Often referred to as the G0 state, quiescence is, however, not a uniform state but with graded depth. Shallow quiescent cells exhibit a higher tendency to revert to proliferation than deep quiescent cells, while deep quiescent cells are still fully reversible under physiological conditions, distinct from senescent cells. Cellular mechanisms underlying the control of quiescence depth and the connection between quiescence and senescence are poorly characterized, representing a missing link in our understanding of tissue homeostasis and regeneration. Here we measured transcriptome changes as rat embryonic fibroblasts moved from shallow to deep quiescence over time in the absence of growth signals. We found that lysosomal gene expression was significantly up-regulated in deep quiescence, and partially compensated for gradually reduced autophagy flux. Reducing lysosomal function drove cells progressively deeper into quiescence and eventually into a senescence-like irreversibly arrested state; increasing lysosomal function, by lowering oxidative stress, progressively pushed cells into shallower quiescence. That is, lysosomal function modulates graded quiescence depth between proliferation and senescence as a dimmer switch. Finally, we found that a gene-expression signature developed by comparing deep and shallow quiescence in fibroblasts can correctly classify a wide array of senescent and aging cell types in vitro and in vivo, suggesting that while quiescence is generally considered to protect cells from irreversible arrest of senescence, quiescence deepening likely represents a common transition path from cell proliferation to senescence, related to aging.

Recent years have witnessed the tendency of measuring a biological sample on multiple omics scales for a comprehensive understanding of how biological activities on varying levels are perturbed by genetic variants, environments, and their interactions. This new trend raises substantial challenges to data integration and fusion, of which the latter is a specific type of integration that applies a uniform method in a scalable manner, to solve biological problems which the multi-omics measurements target. Fusion-based analysis has advanced rapidly in the past decade, thanks to application drivers and theoretical breakthroughs in mathematics, statistics, and computer science. We will briefly address these methods from methodological and mathematical perspectives and categorize them into three types of approaches: data fusion (a narrowed definition as compared to the general data fusion concept), model fusion, and mixed fusion. We will demonstrate at least one typical example in each specific category to exemplify the characteristics, principles, and applications of the methods in general, as well as discuss the gaps and potential issues for future studies.

A chemopreventive effect of aspirin (ASA) on lung cancer risk is supported by epidemiologic and preclinical studies. We conducted a randomized, double-blinded study in current heavy smokers to compare modulating effects of intermittent versus continuous low-dose ASA on nasal epithelium gene expression and arachidonic acid (ARA) metabolism. Fifty-four participants were randomized to intermittent (ASA 81 mg daily for one week/placebo for one week) or continuous (ASA 81 mg daily) for 12 weeks. Low-dose ASA suppressed urinary prostaglandin E2 metabolite (PGEM; change of −4.55 ± 11.52 from baseline 15.44 ± 13.79 ng/mg creatinine for arms combined, P = 0.02), a surrogate of COX-mediated ARA metabolism, but had minimal effects on nasal gene expression of nasal or bronchial gene-expression signatures associated with smoking, lung cancer, and chronic obstructive pulmonary disease. Suppression of urinary PGEM correlated with favorable changes in a smoking-associated gene signature (P < 0.01). Gene set enrichment analysis (GSEA) showed that ASA intervention led to 1,079 enriched gene sets from the Canonical Pathways within the Molecular Signatures Database. In conclusion, low-dose ASA had minimal effects on known carcinogenesis gene signatures in nasal epithelium of current smokers but results in wide-ranging genomic changes in the nasal epithelium, demonstrating utility of nasal brushings as a surrogate to measure gene-expression responses to chemoprevention. PGEM may serve as a marker for smoking-associated gene-expression changes and systemic inflammation. Future studies should focus on NSAIDs or agent combinations with broader inhibition of pro-inflammatory ARA metabolism to shift gene signatures in an anti-carcinogenic direction.

BackgroundIn this era of data science-driven bioinformatics, machine learning research has focused on feature selection as users want more interpretation and post-hoc analyses for biomarker detection. However, when there are more features (i.e., transcript) than samples (i.e., mice or human samples) in a study, this poses major statistical challenges in biomarker detection tasks as traditional statistical techniques are underpowered in high dimension. Second and third order interactions of these features pose a substantial combinatoric dimensional challenge. In computational biology, random forest1 (RF) classifiers are widely used2–7 due to their flexibility, powerful performance, and robustness to “P predictors ≫ subjects N” difficulties and their ability to rank features. We propose binomialRF, a feature selection technique in RFs that provides an alternative interpretation for features using a correlated binomial distribution and scales efficiently to analyze multiway interactions.MethodsbinomialRF treats each tree in a RF as a correlated but exchangeable binary trial. It determines importance by constructing a test statistic based on a feature’s selection frequency to compute its rank, nominal p-value, and multiplicity-adjusted q-value using a one-sided hypothesis test with a correlated binomial distribution. A distributional adjustment addresses the co-dependencies among trees as these trees subsample from the same dataset. The proposed algorithm efficiently identifies multiway nonlinear interactions by generalizing the test statistic to count sub-trees.ResultsIn simulations and in the Madelon benchmark datasets studies, binomialRF showed computational gains (up to 30 to 600 times faster) while maintaining competitive variable precision and recall in identifying biomarkers’ main effects and interactions. In two clinical studies, the binomialRF algorithm prioritizes previously-published relevant pathological molecular mechanisms (features) with high classification precision and recall using features alone, as well as with their statistical interactions alone.ConclusionbinomialRF extends upon previous methods for identifying interpretable features in RFs and brings them together under a correlated binomial distribution to create an efficient hypothesis testing algorithm that identifies biomarkers’ main effects and interactions. Preliminary results in simulations demonstrate computational gains while retaining competitive model selection and classification accuracies. Future work will extend this framework to incorporate ontologies that provide path-way-level feature selection from gene expression input data.AvailabilityGithub: https://github.com/SamirRachidZaim/binomialRFSupplementary informationSupplementary analyses and results are available at https://github.com/SamirRachidZaim/binomialRF_simulationStudy

Due to the rapid development of various social networks, the spatial autoregressive (SAR) model is becoming an important tool in social network analysis. However, major bottlenecks remain in analyzing large-scale networks (e.g., Facebook has over 700 million active users), including computational scalability, estimation consistency, and proper network sampling. To address these challenges, we propose a novel least squares estimator (LSE) for analyzing large sparse networks based on the SAR model. Computationally, the LSE is linear in the network size, making it scalable to analysis of huge networks. In theory, the LSE is $\sqrt{n}$-consistent and asymptotically normal under certain regularity conditions. A new LSE-based network sampling technique is further developed, which can automatically adjust autocorrelation between sampled and unsampled units and hence guarantee valid statistical inferences. Moreover, we generalize the LSE approach for the classical SAR model to more complex networks associated with multiple sources of social interaction effect. Numerical results for simulated and real data are presented to illustrate performance of the LSE.

Background: Developing patient-centric baseline standards that enable the detection of clinically significant outlier gene products on a genome-scale remains an unaddressed challenge required for advancing personalized medicine beyond the small pools of subjects implied by &ldquo, precision medicine&rdquo, This manuscript proposes a novel approach for reference standard development to evaluate the accuracy of single-subject analyses of transcriptomes and offers extensions into proteomes and metabolomes. In evaluation frameworks for which the distributional assumptions of statistical testing imperfectly model genome dynamics of gene products, artefacts and biases are confounded with authentic signals. Model confirmation biases escalate when studies use the same analytical methods in the discovery sets and reference standards. In such studies, replicated biases are confounded with measures of accuracy. We hypothesized that developing method-agnostic reference standards would reduce such replication biases. We propose to evaluate discovery methods with a reference standard derived from a consensus of analytical methods distinct from the discovery one to minimize statistical artefact biases. Our methods involve thresholding effect-size and expression-level filtering of results to improve consensus between analytical methods. We developed and released an R package &ldquo, referenceNof1&rdquo, to facilitate the construction of robust reference standards. Results: Since RNA-Seq data analysis methods often rely on binomial and negative binomial assumptions to non-parametric analyses, the differences create statistical noise and make the reference standards method dependent. In our experimental design, the accuracy of 30 distinct combinations of fold changes (FC) and expression counts (hereinafter &ldquo, expression&rdquo, ) were determined for five types of RNA analyses in two different datasets. This design was applied to two distinct datasets: Breast cancer cell lines and a yeast study with isogenic biological replicates in two experimental conditions. Furthermore, the reference standard (RS) comprised all RNA analytical methods with the exception of the method testing accuracy. To mitigate biases towards a specific analytical method, the pairwise Jaccard Concordance Index between observed results of distinct analytical methods were calculated for optimization. Optimization through thresholding effect-size and expression-level reduced the greatest discordances between distinct methods&rsquo, analytical results and resulted in a 65% increase in concordance. Conclusions: We have demonstrated that comparing accuracies of different single-subject analysis methods for clinical optimization in transcriptomics requires a new evaluation framework. Reliable and robust reference standards, independent of the evaluated method, can be obtained under a limited number of parameter combinations: Fold change (FC) ranges thresholds, expression level cutoffs, and exclusion of the tested method from the RS development process. When applying anticonservative reference standard frameworks (e.g., using the same method for RS development and prediction), most of the concordant signal between prediction and Gold Standard (GS) cannot be confirmed by other methods, which we conclude as biased results. Statistical tests to determine DEGs from a single-subject study generate many biased results requiring subsequent filtering to increase reliability. Conventional single-subject studies pertain to one or a few patient&rsquo, s measures over time and require a substantial conceptual framework extension to address the numerous measures in genome-wide analyses of gene products. The proposed referenceNof1 framework addresses some of the inherent challenges for improving transcriptome scale single-subject analyses by providing a robust approach to constructing reference standards.

We propose a new binary classification and variable selection technique especially designed for high-dimensional predictors. Among many predictors, typically, only a small fraction of them have significant impact on prediction. In such a situation, more interpretable models with better prediction accuracy can be obtained by variable selection along with classification. By adding an l1-type penalty to the loss function, common classification methods such as logistic regression or support vector machines (SVM) can perform variable selection. Existing penalized SVM methods all attempt to jointly solve all the parameters involved in the penalization problem altogether. When data dimension is very high, the joint optimization problem is very complex and involves a lot of memory allocation. In this article, we propose a new penalized forward search technique that can reduce high-dimensional optimization problems to one-dimensional optimization by iterating the selection steps. The new algorithm can be regarded ...

SUMMARY In modern experiments, functional and nonfunctional data are often encountered simultaneously when observations are sampled from random processes and high-dimensional scalar covariates. It is difficult to apply existing methods for model selection and estimation. We propose a new class of partially functional linear models to characterize the regression between a scalar response and covariates of both functional and scalar types. The new approach provides a unified and flexible framework that simultaneously takes into account multiple functional and ultrahigh-dimensional scalar predictors, enables us to identify important features, and offers improved interpretability of the estimators. The underlying processes of the functional predictors are considered to be infinite-dimensional, and one of our contributions is to characterize the effects of regularization on the resulting estimators. We establish the consistency and oracle properties of the proposed method under mild conditions, demonstrate its performance with simulation studies, and illustrate its application using air pollution data.

A large body of epidemiologic evidence has shown that use of progestin-containing preparations lowers ovarian cancer risk. The purpose of the current study was to gather further preclinical evidence supporting progestins as cancer chemopreventives by demonstrating progestin-activation of surrogate endpoint biomarkers pertinent to cancer prevention in the genital tract of women at increased risk of ovarian cancer. There were 64 women enrolled in a multi-institutional randomized trial who chose to undergo risk-reducing bilateral salpingo-oophorectomy (BSO) and to receive the progestin levonorgestrel or placebo for 4 to 6 weeks prior to undergoing BSO. The ovarian and fallopian tube epithelia (FTE) were compared immunohistochemically for effects of levonorgestrel on apoptosis (primary endpoint). Secondary endpoints included TGFβ isoform expression, proliferation, and karyometric features of nuclear abnormality. In both the ovary and fallopian tube, levonorgestrel did not confer significant changes in apoptosis or expression of the TGFβ1, 2, or 3 isoforms. In the ovarian epithelium, treatment with levonorgestrel significantly decreased the proliferation index. The mean ovarian Ki-67 value in the placebo arm was 2.027 per 100 cells versus 0.775 per 100 cells in the levonorgestrel arm (two-sided P value via Mann–Whitney U test = 0.0114). The karyometric signature of nuclei in both the ovarian and FTE deviated significantly from normal controls (women at average risk of ovarian cancer), but was significantly less abnormal in women treated with levonorgestrel. These karyometric data further support the idea that progestins may clear genetically abnormal cells and act as chemopreventive agents against ovarian and fallopian tube cancer.

Numerous physiological and pathological phenomena are associated with the quiescent state of a cell. Cellular quiescence is a heterogeneous resting state; cells in deep than shallow quiescence require stronger growth stimulation to exit quiescence and reenter the cell cycle. Despite the importance of quiescent cells such as stem and progenitor cells to tissue homeostasis and repair, cellular mechanisms controlling the depth of cellular quiescence are poorly understood. Here we began by analyzing transcriptome changes as rat embryonic fibroblasts moved progressively deeper into quiescence under increasingly longer periods of serum starvation. We found that lysosomal gene expression was significantly upregulated in deep than shallow quiescence, which compensated for gradually reduced autophagy flux observed during quiescence deepening. Consistently, we show that inhibiting lysosomal function drove cells deeper into quiescence and eventually into a senescence-like irreversibly arrested state. By contrast, increasing lysosomal function progressively pushed cells into shallower quiescence. That is, lysosomal function modulates quiescence depth continuously like a dimmer switch. Mechanistically, we show that lysosomal function prevents quiescence deepening by reducing oxidative stress in the cell. Lastly, we show that a gene expression signature developed by comparing deep and shallow quiescent cells can correctly classify senescent and aging cells in a wide array of cell lines in vitro and tissues in vivo, suggesting that quiescence deepening, senescence, and aging may share common regulatory mechanisms.

Quadratic regression (QR) models naturally extend linear models by considering interaction effects between the covariates. To conduct model selection in QR, it is important to maintain the hierarchical model structure between main effects and interaction effects. Existing regularization methods generally achieve this goal by solving complex optimization problems, which usually demands high computational cost and hence are not feasible for high-dimensional data. This article focuses on scalable regularization methods for model selection in high-dimensional QR. We first consider two-stage regularization methods and establish theoretical properties of the two-stage LASSO. Then, a new regularization method, called regularization algorithm under marginality principle (RAMP), is proposed to compute a hierarchy-preserving regularization solution path efficiently. Both methods are further extended to solve generalized QR models. Numerical results are also shown to demonstrate performance of the methods. Supplementary materials for this article are available online.

In clinical studies with time-to-event as a primary endpoint, one main interest is to find the best treatment strategy to maximize patients’ mean survival time. Due to patient’s heterogeneity in response to treatments, great efforts have been devoted to developing optimal treatment regimes by integrating individuals’ clinical and genetic information. A main challenge arises in the selection of important variables that can help to build reliable and interpretable optimal treatment regimes since the dimension of predictors may be high. In this paper, we propose a robust loss-based estimation framework that can be easily coupled with shrinkage penalties for both estimation of optimal treatment regimes and variable selection. The asymptotic properties of the proposed estimators are studied. Moreover, a model-free estimator of restricted mean survival time under the derived optimal treatment regime is developed and its asymptotic property is studied. Simulations are conducted to assess the empirical performance of the proposed method for parameter estimation, variable selection, and optimal treatment decision. An application to an AIDS clinical trial data set is given to illustrate the method.

Different from the standard treatment discovery framework which is used for finding single treatments for a homogenous group of patients, personalized medicine involves finding therapies that are tailored to each individual in a heterogeneous group. In this paper, we propose a new semiparametric additive single-index model for estimating individualized treatment strategy. The model assumes a flexible and nonparametric link function for the interaction between treatment and predictive covariates. We estimate the rule via monotone B-splines and establish the asymptotic properties of the estimators. Both simulations and an real data application demonstrate that the proposed method has a competitive performance.

The concept of P-value was proposed by Fisher to measure inconsistency of data with a specified null hypothesis, and it plays a central role in statistical inference. For classical linear regression analysis, it is a standard procedure to calculate P-values for regression coefficients based on least squares estimator (LSE) to determine their significance. However, for high dimensional data when the number of predictors exceeds the sample size, ordinary least squares are no longer proper and there is not a valid definition for P-values based on LSE. It is also challenging to define sensible P-values for other high dimensional regression methods such as penalization and resampling methods. In this paper, we introduce a new concept called oracle P-value to generalize traditional P-values based on LSE to high dimensional sparse regression models. Then we propose several estimation procedures to approximate oracle P-values for real data analysis. We show that the oracle P-value framework is useful for developing new and powerful tools to enhance high dimensional data analysis, including variable ranking, variable selection, and screening procedures with false discovery rate (FDR) control. Numerical examples are then presented to demonstrate performance of the proposed methods.

Summary In high-dimensional data analysis, it is of primary interest to reduce the data dimensionality without loss of information. Sufficient dimension reduction (SDR) arises in this context, and many successful SDR methods have been developed since the introduction of sliced inverse regression (SIR) [Li (1991) Journal of the American Statistical Association 86, 316–327]. Despite their fast progress, though, most existing methods target on regression problems with a continuous response. For binary classification problems, SIR suffers the limitation of estimating at most one direction since only two slices are available. In this article, we develop a new and flexible probability-enhanced SDR method for binary classification problems by using the weighted support vector machine (WSVM). The key idea is to slice the data based on conditional class probabilities of observations rather than their binary responses. We first show that the central subspace based on the conditional class probability is the same as that based on the binary response. This important result justifies the proposed slicing scheme from a theoretical perspective and assures no information loss. In practice, the true conditional class probability is generally not available, and the problem of probability estimation can be challenging for data with large-dimensional inputs. We observe that, in order to implement the new slicing scheme, one does not need exact probability values and the only required information is the relative order of probability values. Motivated by this fact, our new SDR procedure bypasses the probability estimation step and employs the WSVM to directly estimate the order of probability values, based on which the slicing is performed. The performance of the proposed probability-enhanced SDR scheme is evaluated by both simulated and real data examples.

The support vector machine (SVM) is a popular learning method for binary classification. Standard SVMs treat all the data points equally, but in some practical problems it is more natural to assign different weights to observations from different classes. This leads to a broader class of learning, the so-called weighted SVMs (WSVMs), and one of their important applications is to estimate class probabilities besides learning the classification boundary. There are two parameters associated with the WSVM optimization problem: one is the regularization parameter and the other is the weight parameter. In this paper we first establish that the WSVM solutions are jointly piecewise-linear with respect to both the regularization and weight parameter. We then develop a state-of-the-art algorithm that can compute the entire trajectory of the WSVM solutions for every pair of the regularization parameter and the weight parameter, at a feasible computational cost. The derived two-dimensional solution surface provides theoretical insight on the behavior of the WSVM solutions. Numerically, the algorithm can greatly facilitate the implementation of the WSVM and automate the selection process of the optimal regularization parameter. We illustrate the new algorithm on various examples.

Risk classification and survival probability prediction are two major goals in survival data analysis because they play an important role in patients' risk stratification, long-term diagnosis, and treatment selection. In this article, we propose a new model-free machine learning framework for risk classification and survival probability prediction based on weighted support vector machines. The new procedure does not require any specific parametric or semiparametric model assumption on data and is therefore capable of capturing non-linear covariate effects. We use numerous simulation examples to demonstrate finite sample performance of the proposed method under various settings. Applications to a glioma tumour data and a breast cancer gene-expression survival data are shown to illustrate the new methodology in real data analysis. Copyright © 2014 John Wiley & Sons, Ltd.

Model selection and estimation are crucial parts of econometrics. This paper introduces a new technique that can simultaneously estimate and select the model in generalized method of moments (GMM) context. The GMM is particularly powerful for analyzing complex data sets such as longitudinal and panel data, and it has wide applications in econometrics. This paper extends the least squares based adaptive elastic net estimator of Zou and Zhang (2009) to nonlinear equation systems with endogenous variables. The extension is not trivial and involves a new proof technique due to estimators lack of closed form solutions. Compared to Bridge-GMM of Caner (2009), we allow for the number of parameters to diverge to infinity as well as collinearity among a large number of variables, also the redundant parameters set to zero via a data dependent technique. This method has the oracle property, meaning that we can estimate nonzero parameters with their standard limit and the redundant parameters are dropped from the equations simultaneously. Numerical examples are used to illustrate the performance of the new method.

Motivation Understanding dynamic, patient-level transcriptomic response to therapy is an important step forward for precision medicine. However, conventional transcriptome analysis aims to discover cohort-level change, lacking the capacity to unveil patient-specific response to therapy. To address this gap, we previously developed two N-of-1-pathways methods, Wilcoxon and Mahalanobis distance, to detect unidirectionally responsive transcripts within a pathway using a pair of samples from a single subject. Yet, these methods cannot recognize bidirectionally (up and down) responsive pathways. Further, our previous approaches have not been assessed in presence of background noise and are not designed to identify differentially expressed mRNAs between two samples of a patient taken in different contexts (e.g. cancer vs non cancer), which we termed responsive transcripts (RTs). Methods We propose a new N-of-1-pathways method, k-Means Enrichment (kMEn), that detects bidirectionally responsive pathways, despite background noise, using a pair of transcriptomes from a single patient. kMEn identifies transcripts responsive to the stimulus through k-means clustering and then tests for an over-representation of the responsive genes within each pathway. The pathways identified by kMEn are mechanistically interpretable pathways significantly responding to a stimulus. Results In ∼9000 simulations varying six parameters, superior performance of kMEn over previous single-subject methods is evident by: (i) improved precision-recall at various levels of bidirectional response and (ii) lower rates of false positives (1-specificity) when more than 10% of genes in the genome are differentially expressed (background noise). In a clinical proof-of-concept, personal treatment-specific pathways identified by kMEn correlate with therapeutic response (p-value Conclusion Through improved single-subject transcriptome dynamics of bidirectionally-regulated signals, kMEn provides a novel approach to identify mechanism-level biomarkers.

Key Points Question Can machine-learning approaches predict opioid overdose risk among fee-for-service Medicare beneficiaries? Findings In this prognostic study of the administrative claims data of 560 057 Medicare beneficiaries, the deep neural network and gradient boosting machine models outperformed other methods for identifying risk, although positive predictive values were low given the low prevalence of overdose episodes. Meaning Machine-learning algorithms using administrative data appear to be a valuable and feasible tool for more accurate identification of opioid overdose risk., Importance Current approaches to identifying individuals at high risk for opioid overdose target many patients who are not truly at high risk. Objective To develop and validate a machine-learning algorithm to predict opioid overdose risk among Medicare beneficiaries with at least 1 opioid prescription. Design, Setting, and Participants A prognostic study was conducted between September 1, 2017, and December 31, 2018. Participants (n = 560 057) included fee-for-service Medicare beneficiaries without cancer who filled 1 or more opioid prescriptions from January 1, 2011, to December 31, 2015. Beneficiaries were randomly and equally divided into training, testing, and validation samples. Exposures Potential predictors (n = 268), including sociodemographics, health status, patterns of opioid use, and practitioner-level and regional-level factors, were measured in 3-month windows, starting 3 months before initiating opioids until loss of follow-up or the end of observation. Main Outcomes and Measures Opioid overdose episodes from inpatient and emergency department claims were identified. Multivariate logistic regression (MLR), least absolute shrinkage and selection operator–type regression (LASSO), random forest (RF), gradient boosting machine (GBM), and deep neural network (DNN) were applied to predict overdose risk in the subsequent 3 months after initiation of treatment with prescription opioids. Prediction performance was assessed using the C statistic and other metrics (eg, sensitivity, specificity, and number needed to evaluate [NNE] to identify one overdose). The Youden index was used to identify the optimized threshold of predicted score that balanced sensitivity and specificity. Results Beneficiaries in the training (n = 186 686), testing (n = 186 685), and validation (n = 186 686) samples had similar characteristics (mean [SD] age of 68.0 [14.5] years, and approximately 63% were female, 82% were white, 35% had disabilities, 41% were dual eligible, and 0.60% had at least 1 overdose episode). In the validation sample, the DNN (C statistic = 0.91; 95% CI, 0.88-0.93) and GBM (C statistic = 0.90; 95% CI, 0.87-0.94) algorithms outperformed the LASSO (C statistic = 0.84; 95% CI, 0.80-0.89), RF (C statistic = 0.80; 95% CI, 0.75-0.84), and MLR (C statistic = 0.75; 95% CI, 0.69-0.80) methods for predicting opioid overdose. At the optimized sensitivity and specificity, DNN had a sensitivity of 92.3%, specificity of 75.7%, NNE of 542, positive predictive value of 0.18%, and negative predictive value of 99.9%. The DNN classified patients into low-risk (76.2% [142 180] of the cohort), medium-risk (18.6% [34 579] of the cohort), and high-risk (5.2% [9747] of the cohort) subgroups, with only 1 in 10 000 in the low-risk subgroup having an overdose episode. More than 90% of overdose episodes occurred in the high-risk and medium-risk subgroups, although positive predictive values were low, given the rare overdose outcome. Conclusions and Relevance Machine-learning algorithms appear to perform well for risk prediction and stratification of opioid overdose, especially in identifying low-risk subgroups that have minimal risk of overdose., This prognostic study evaluates the use of machine-learning methods, with prescription drug and claims data, in detecting opioid overdose risk in Medicare beneficiaries with at least 1 opioid prescription.

High dimensional data are nowadays encountered in various branches of science. Variable selection techniques play a key role in analyzing high dimensional data. Generally two approaches for variable selection in the high dimensional data setting are considered—forward selection methods and penalization methods. In the former, variables are introduced in the model one at a time depending on their ability to explain variation and the procedure is terminated at some stage following some stopping rule. In penalization techniques such as the least absolute selection and shrinkage operator (LASSO), as optimization procedure is carried out with an added carefully chosen penalty function, so that the solutions have a sparse structure. Recently, the idea of penalized forward selection has been introduced. The motivation comes from the fact that the penalization techniques like the LASSO give rise to closed form expressions when used in one dimension, just like the least square estimator. Hence one can repeat such a procedure in a forward selection setting until it converges. The resulting procedure selects sparser models than comparable methods without compromising on predictive power. However, when the regressor is high dimensional, it is typical that many predictors are highly correlated. We show that in such situations, it is possible to improve stability and computational efficiency of the procedure further by introducing an orthogonalization step. At each selection step, variables potentially available to be selected in the model are screened on the basis of their correlation with variables already in the model, thus preventing unnecessary duplication. The new strategy, called the Selection Technique in Orthogonalized Regression Models (STORM), turns out to be extremely successful in reducing the model dimension further and also leads to improved predicting power. We also consider an aggressive version of the STORM, where a potential predictor will be permanently removed from further consideration if its regression coefficient is estimated as zero at any stage. We shall carry out a detailed simulation study to compare the newly proposed method with existing ones and analyze a real dataset.  2013 Wiley Periodicals, Inc. Statistical Analysis and Data Mining 6: 557-564, 2013