Search

We introduce methods for signal and associated variability estimation based on hierarchical nonparametric smoothing with application to the Sleep Heart Health Study (SHHS). SHHS is the largest electroencephalographic (EEG) collection of sleep-related data, which contains, at each visit, two quasi-continuous EEG signals for each subject. The signal features extracted from EEG data are then used in second level analyses to investigate the relation between health, behavioral, or biometric outcomes and sleep. Using subject specific signals estimated with known variability in a second level regression becomes a nonstandard measurement error problem. We propose and implement methods that take into account cross-sectional and longitudinal measurement error. The research presented here forms the basis for EEG signal processing for the SHHS.

We introduce Generalized Multilevel Functional Linear Models (GMFLMs), a novel statistical framework for regression models where exposure has a multilevel functional structure. We show that GMFLMs are, in fact, generalized multilevel mixed models (GLMMs). Thus, GMFLMs can be analyzed using the mixed effects inferential machinery and can be generalized within a well researched statistical framework. We propose and compare two methods for inference: 1) a two-stage frequentist approach; and 2) a joint Bayesian analysis. Our methods are motivated by and applied to the Sleep Heart Health Study (SHHS), the largest community cohort study of sleep. However, our methods are general and easy to apply to a wide spectrum of emerging biological and medical data sets. Supplemental materials for this article are available online.

In this article we are concerned with functional imaging of the colon to assess the kinetics of microbicide lubricants. The overarching goal is to understand the distribution of the lubricants in the colon. Such information is crucial for understanding the potential impact of microbicides on human immunodeficiency virus transmission. The experiment was conducted by imaging a radiolabeled lubricant distributed in the subject’s colon. The tracer imaging was conducted via single-photon emission computed tomography, a noninvasive, in vivo functional imaging technique. We have developed a novel principal curve algorithm to construct a three-dimensional curve through the colon images. The algorithm was tested and debugged on several difficult two-dimensional images of familiar curves where the original principal curve algorithm does not apply. The final curve fit to the colon data is compared with experimental sigmoidoscope collection.

The goal of our article is to provide a transparent, robust, and computationally feasible statistical platform for restricted likelihood ratio testing (RLRT) for zero variance components in linear mixed models. This problem is nonstandard because under the null hypothesis the parameter is on the boundary of the parameter space. Our proposed approach is different from the asymptotic results of Stram and Lee who assumed that the outcome vector can be partitioned into many independent subvectors. Thus, our methodology applies to a wider class of mixed models, which includes models with a moderate number of clusters or nonparametric smoothing components. We propose two approximations to the finite sample null distribution of the RLRT statistic. Both approximations converge weakly to the asymptotic distribution obtained by Stram and Lee when their assumptions hold. When their assumptions do not hold, we show in extensive simulation studies that both approximations outperform the Stram and Lee approximation and...

A new method (Event-Related Causality, ERC) is proposed for the investigation of functional interactions between brain regions during cognitive processing. ERC estimates the direction, intensity, spectral content, and temporal course of brain activity propagation within a cortical network. ERC is based upon the short-time directed transfer function (SDTF), which is measured in short EEG epochs during multiple trials of a cognitive task, as well as the direct directed transfer function (dDTF), which distinguishes direct interactions between brain regions from indirect interactions via brain regions. ERC uses new statistical methods for comparing estimates of causal interactions during prestimulus ''baseline'' epochs and during poststimulus ''activated'' epochs in order to estimate event-related increases and decreases in the functional interactions between cortical network components during cog- nitive tasks. The utility of the ERC approach is demonstrated through its application to human electro- corticographic recordings (ECoG) of a simple language task. ERC analyses of these ECoG recordings reveal frequency-dependent interactions, particularly in high gamma (>60 Hz) frequencies, between brain regions known to participate in the recorded language task, and the temporal evolution of these interactions is consistent with the putative processing stages of this task. The method may be a useful tool for investigating the dynamics of causal interactions between various brain regions during cogni- tive task performance. Hum Brain Mapp 29:1170-1192, 2008. V C 2007 Wiley-Liss, Inc.

A mathematical model of the electrocatalytic hydrogenation of phenol on a Pd/Al 2 O 3 (10%, w/w) catalyst is developed and presented from the chemical kinetic to the fluid mechanical point of view in a laboratory scale cell. The model includes the heterogenous electrochemical steps at the microscopic scale, integrated in a full transient, three-dimensional representation of the fluid mechanics and species concentration in the electrochemical cell. In this reactor, the catalytic particles are dispersed and recirculated in the solution. The brief contact of the powder with a porous reticulated vitreous carbon is sufficient to obtain adsorbed hydrogen on the nanoaggregates who contains the phenol adsorbed. The isotherm of Langmuir and the pseudo-second-order model, based on assumption of second-order mechanism are developed. The aim of this work is to understand and validate the pseudo-second-order adsorption model when used with hydrogen adsorption isotherm, and to develop a full model of the electrochemical cell. The proposed model could be used to improve the design of industrial scale reactors. The model is validated using experimental results from [A. Bannari, C. Cirtiu, F. Kerdouss, P. Proulx, H. Menard, Chem. Eng. Process. 45 (2006) 471].

We present a state-of-the-art application of smoothing for dependent bivariate binomial spatial data to Loa loa prevalence mapping in West Africa. This application starts with the nonspatial calibration of survey instruments, continues with the spatial model building and assessment, and ends with robust, tested software intended for use by field workers for online prevalence map updating. From a statistical perspective, we address several important methodological issues: building spatial models that are sufficiently complex to capture the structure of the data but remain computationally usable, reducing the computational burden in the handling of very large covariate data sets, and devising methods for comparing spatial prediction methods for a given exceedance policy threshold.

Magnetic resonance imaging (MRI) intensities are acquired in arbitrary units, making scans non-comparable across sites and between subjects. Intensity normalization is a first step for the improvement of comparability of the images across subjects. However, we show that unwanted inter-scan variability associated with imaging site, scanner effect, and other technical artifacts is still present after standard intensity normalization in large multi-site neuroimaging studies. We propose RAVEL (Removal of Artificial Voxel Effect by Linear regression), a tool to remove residual technical variability after intensity normalization. As proposed by SVA and RUV [Leek and Storey, 2007, 2008, Gagnon-Bartsch and Speed, 2012], two batch effect correction tools largely used in genomics, we decompose the voxel intensities of images registered to a template into a biological component and an unwanted variation component. The unwanted variation component is estimated from a control region obtained from the cerebrospinal fluid (CSF), where intensities are known to be unassociated with disease status and other clinical covariates. We perform a singular value decomposition (SVD) of the control voxels to estimate factors of unwanted variation. We then estimate the unwanted factors using linear regression for every voxel of the brain and take the residuals as the RAVEL-corrected intensities. We assess the performance of RAVEL using T1-weighted (T1-w) images from more than 900 subjects with Alzheimer's disease (AD) and mild cognitive impairment (MCI), as well as healthy controls from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. We compare RAVEL to two intensity-normalization-only methods: histogram matching and White Stripe. We show that RAVEL performs best at improving the replicability of the brain regions that are empirically found to be most associated with AD, and that these regions are significantly more present in structures impacted by AD (hippocampus, amygdala, parahippocampal gyrus, enthorinal area, and fornix stria terminals). In addition, we show that the RAVEL-corrected intensities have the best performance in distinguishing between MCI subjects and healthy subjects using the mean hippocampal intensity (AUC=67%), a marked improvement compared to results from intensity normalization alone (AUC=63% and 59% for histogram matching and White Stripe, respectively). RAVEL is promising for many other imaging modalities.

STUDY OBJECTIVES The current study sought to determine whether sleep duration and change in sleep duration are associated with all-cause mortality in a community sample of middle-aged and older adults while accounting for several confounding factors including prevalent sleep-disordered breathing (SDB). METHODS Habitual sleep duration was assessed using self-report (< 7, 7-8, ≥ 9 h/night) at the baseline and at the follow-up visits of the Sleep Heart Health Study. Techniques of survival analysis were used to relate habitual sleep duration and change in sleep duration to all-cause mortality after adjusting for covariates such as age, sex, race, body mass index, smoking history, prevalent hypertension, diabetes, cardiovascular disease, antidepressant medication use, and SDB severity. RESULTS Compared to a sleep duration of 7-8 h/night, habitually long sleep duration (≥ 9 h/night), but not short sleep duration (< 7 h/night), was associated with all-cause mortality with an adjusted hazards ratio of 1.25 (95% confidence interval [CI]: 1.05, 1.47). Participants who progressed from short or normal sleep duration to long sleep duration had increased risk for all-cause mortality with adjusted hazard ratios of 1.75 (95% CI: 1.08, 2.78) and 1.63 (95% CI: 1.26, 2.13), respectively. Finally, a change from long to short sleep duration was also associated with all-cause mortality. CONCLUSION Long sleep duration or a shift from long to short sleep duration are independently associated with all-cause mortality.

BACKGROUND: Data collected by wearable accelerometry devices can be used to identify periods of sustained harmonic walking. This report aims to establish whether the features of walking identified in the laboratory and free-living environments are associated with each other as well as measures of physical function, mobility, fatigability, and fitness. METHODS: Fifty-one older adults (mean age 78.31) enrolled in the Developmental Epidemiologic Cohort Study were included in the analyses. The study included an “in-the-lab” component as well as 7 days of monitoring “in-the-wild” (free living). Participants were equipped with hip-worn Actigraph GT3X+ activity monitors, which collect raw accelerometry data. We applied a walking identification algorithm and defined features of walking, including participant-specific walking acceleration and cadence. The association between these walking features and physical function, mobility, fatigability, and fitness was quantified using linear regression analysis. RESULTS: Acceleration and cadence estimated from “in-the-lab” and “in-the-wild” data were significantly associated with each other (p < .05). However, walking acceleration “in-the-lab” was on average 96% higher than “in-the-wild,” whereas cadence “in-the-lab” was on average 20% higher than “in-the-wild.” Acceleration and cadence were associated with measures of physical function, mobility, fatigability, and fitness (p < .05) in both “in-the-lab” and “in-the-wild” settings. In addition, “in-the-wild” daily walking time was associated with fitness (p < .05). CONCLUSIONS: The quantitative difference in proposed walking features indicates that participants may overperform when observed “in-the-lab.” Also, proposed features of walking were significantly associated with measures of physical function, mobility, fatigability, and fitness, which provides evidence of convergent validity.

Haploinsufficiency of the SYNGAP1 gene, which codes for a Ras GTPase-activating protein, impairs cognition both in humans and in mice. Decrease of Syngap1 in mice has been previously shown to cause cognitive deficits at least in part by inducing alterations in glutamatergic neurotransmission and premature maturation of excitatory connections. Whether Syngap1 plays a role in the development of cortical GABAergic connectivity and function remains unclear. Here, we show that Syngap1 haploinsufficiency significantly reduces the formation of perisomatic innervations by parvalbumin-positive basket cells, a major population of GABAergic neurons, in a cell-autonomous manner. We further show that Syngap1 haploinsufficiency in GABAergic cells derived from the medial ganglionic eminence impairs their connectivity, reduces inhibitory synaptic activity and cortical gamma oscillation power, and causes cognitive deficits. Our results indicate that Syngap1 plays a critical role in GABAergic circuit function and further suggest that Syngap1 haploinsufficiency in GABAergic circuits may contribute to cognitive deficits., Glutamatergic signalling regulation by Syngap1 has been linked to intellectual disabilities. Here, the authors find Syngap1 also regulates cortical GABAergic synaptic signalling development and that this reduced inhibitory signalling contributes to cognitive deficits in a mouse model.

Accelerometers have been widely deployed in public health studies in recent years. While they collect high-resolution acceleration signals (e.g., 10-100 Hz), research has mainly focused on summarized metrics provided by accelerometers manufactures, such as the activity count (AC) by ActiGraph or Actical. Such measures do not have a publicly available formula, lack a straightforward interpretation, and can vary by software implementation or hardware type. To address these problems, we propose the physical activity index (AI), a new metric for summarizing raw tri-axial accelerometry data. We compared this metric with the AC and another recently proposed metric for raw data, Euclidean Norm Minus One (ENMO), against energy expenditure. The comparison was conducted using data from the Objective Physical Activity and Cardiovascular Health Study, in which 194 women 60-91 years performed 9 lifestyle activities in the laboratory, wearing a tri-axial accelerometer (ActiGraph GT3X+) on the hip set to 30 Hz and an Oxycon portable calorimeter, to record both tri-axial acceleration time series (converted into AI, AC, and ENMO) and oxygen uptake during each activity (converted into metabolic equivalents (METs)) at the same time. Receiver operating characteristic analyses indicated that both AI and ENMO were more sensitive to moderate and vigorous physical activities than AC, while AI was more sensitive to sedentary and light activities than ENMO. AI had the highest coefficients of determination for METs (0.72) and was a better classifier of physical activity intensity than both AC (for all intensity levels) and ENMO (for sedentary and light intensity). The proposed AI provides a novel and transparent way to summarize densely sampled raw accelerometry data, and may serve as an alternative to AC. The AI's largely improved sensitivity on sedentary and light activities over AC and ENMO further demonstrate its advantage in studies with older adults.

The electrocatalytic hydrogenation (ECH) of octyl aldehyde (octanal) to octyl alcohol (octan-1-ol) was investigated using commercial Pd/alumina catalysts in aqueous ethanol. The influence of different parameters, such as catalyst support, current intensity, polarity of solvent, supporting electrolyte, and octanal concentration, on the electrocatalytic hydrogenation of octanal was studied.Key words: octanal, octan-1-ol, Pd/alumina catalyst, electrocatalytic hydrogenation.

The null hypothesis-based statistics CUSUM and QS are widely used for testing parameter stability. We provide examples, extensive simulation studies and theoretical results showing that these statistics fail to detect obvious shifts in the mean of a time series. Moreover, the detection probability can decrease when the magnitude of the shift in mean increases. Estimation of nuisance parameters under the null is identified as an important cause of this counterintuitive behavior of the power function. Results indicate that tests designed for the specific alternative of a shift in mean of a time series should be preferred. §This paper was presented at the 2002 Winter Meetings of the Econometric Society in Atlanta, GA.

Penalized splines have become an increasingly popular tool for nonparametric smoothing because of their use of low-rank spline bases, which makes computations tractable while maintaining accuracy as good as smoothing splines. This article extends penalized spline methodology by both modeling the variance function nonparametrically and using a spatially adaptive smoothing parameter. This combination is needed for satisfactory inference and can be implemented effectively by Bayesian MCMC. The variance process controlling the spatially adaptive shrinkage of the mean and the variance of the heteroscedastic error process are modeled as log-penalized splines. We discuss the choice of priors and extensions of the methodology, in particular, to multivariate smoothing. A fully Bayesian approach provides the joint posterior distribution of all parameters, in particular, of the error standard deviation and penalty functions. MATLAB, C, and FORTRAN programs implementing our methodology are publicly available.

Catalytic and electrocatalytic hydrogenation of benzophenone was performed in aqueous ethanol onto 10% (w/w) Pd–alumina supported catalyst and finely divided Pd catalyst. Efficiency of catalytic and electrocatalytic hydrogenation of the studied ketone is discussed here as a comparative study.

Reduced kidney function is a risk factor for cardiovascular morbidity and mortality, and both heart failure (HF) and kidney failure incidences are increasing. This study therefore sought to determine the effect of decreased kidney function on HF incidence in a population-based study of middle-aged adults. From 1987 through 2002, 14,857 participants of the Atherosclerosis Risk in Communities (ARIC) study who were free of prevalent HF at baseline were followed for incident HF hospitalization or death (International Classification of Diseases, Ninth Revision/10th Revision 428/I50). Estimated GFR (eGFR) was calculated using the abbreviated Modification of Diet in Renal Disease (MDRD) Study equation, and kidney function was categorized as normal (eGFRor =90 ml/min per 1.73 m(2); n = 7143), mildly reduced (eGFR 60 to 89 ml/min per 1.73 m(2); n = 7311), and moderately/severely reduced (eGFR60 ml/min per 1.73 m(2); n = 403). Cox proportional hazards models were used to control for demographic and cardiovascular risk factors; analyses were stratified by the presence of coronary heart disease at baseline. During a mean follow-up of 13.2 yr, 1193 participants developed HF. The incidence of HF was three-fold higher for individuals with eGFR60 ml/min per 1.73 m(2) compared to the reference group with eGFRor =90 ml/min per 1.73 m(2) (18 versus 6 per 1000 person-years). The overall adjusted relative hazard of developing HF was 1.94 (1.49 to 2.53) for individuals with eGFR60 ml/min per 1.73 m(2) compared to the reference group and was significantly increased for individuals with and without prevalent coronary heart disease at baseline. A substantially greater decline in kidney function occurred in individuals concomitant with HF hospitalization/death compared to those who did not develop HF. In summary, middle-aged adults with moderately/severely reduced kidney function are at high risk for developing HF.

Biomonitoring for manganese (Mn) exposure is important due to its potential to cause adverse health effects. In this study, we investigate how different sample preparation methods (simple dilution, digestion, volumetric, gravimetric), calibration protocols (aqueous, blood-based, standard additions), and instrumental techniques affect Mn method bias and analytical imprecision. The techniques used included graphite furnace atomic absorption spectrometry (GFAAS), dynamic reaction cell inductively coupled plasma mass spectrometry (DRC-ICP-MS), and sector field (SF-) ICP-MS. We analyzed NIST SRM 1643e Trace Elements in Water and SRM 1598a Inorganic Constituents in Animal Serum (both certified for Mn), and SRM 955c Toxic Metals in Caprine Blood – Level 1 (not certified for Mn). Various matrix effects in ICP-MS produced inaccurate results for SRM 1643e and discrepant results for SRM 955c. In the absence of a certified value for Mn in SRM 955c, we assigned a “consensus” value by combining data from the New York State Department of Health (NYS), the Centers for Disease Control and Prevention (CDC) and the Centre de toxicologie du Québec (CTQ). With this interlaboratory approach, we established an “all-lab” consensus value of 16.3 ± 0.8 μg L−1 based on data from DRC-ICP-MS with simple dilution sample preparation and blood-based calibration. We also assigned an “all-method” consensus value of 16.3 ± 0.9 μg L−1 based on GFAAS and SF-ICP-MS data from the NYS lab and the DRC-ICP-MS all-lab consensus value. Although the expanded uncertainty (U) calculated for the consensus values may not fully account for all sources of uncertainty, it does show the relative variation that might be expected from one study to the next for the determination of Mn in blood.

Background and Purpose— The location of intracerebral hemorrhage (ICH) is currently described in a qualitative way; we provide a quantitative framework for estimating ICH engagement and its relevance to stroke outcomes. Methods— We analyzed 111 patients with ICH from the Minimally Invasive Surgery Plus Recombinant-Tissue Plasminogen Activator for Intracerebral Evacuation (MISTIE) II clinical trial. We estimated ICH engagement at a population level using image registration of computed tomographic scans to a template and a previously labeled atlas. Predictive regions of National Institutes of Health Stroke Scale and Glasgow Coma Scale stroke severity scores, collected at enrollment, were estimated. Results— The percent coverage of the ICH by these regions strongly outperformed the reader-labeled locations. The adjusted R 2 almost doubled from 0.129 (reader-labeled model) to 0.254 (quantitative location model) for National Institutes of Health Stroke Scale and more than tripled from 0.069 (reader-labeled model) to 0.214 (quantitative location model). A permutation test confirmed that the new predictive regions are more predictive than chance: P< 0.001 for National Institutes of Health Stroke Scale and P< 0.01 for Glasgow Coma Scale. Conclusions— Objective measures of ICH location and engagement using advanced computed tomographic imaging processing provide finer, objective, and more quantitative anatomic information than that provided by human readers. Clinical Trial Registration— URL: http://www.clinicaltrials.gov . Unique identifier: NCT00224770.

We propose fast and scalable statistical methods for the analysis of hundreds or thousands of high dimensional vectors observed at multiple visits. The proposed inferential methods do not require loading the entire data set at once in the computer memory and instead use only sequential access to data. This allows deployment of our methodology on low-resource computers where computations can be done in minutes on extremely large data sets. Our methods are motivated by and applied to a study where hundreds of subjects were scanned using Magnetic Resonance Imaging (MRI) at two visits roughly five years apart. The original data possesses over ten billion measurements. The approach can be applied to any type of study where data can be unfolded into a long vector including densely observed functions and images. Supplemental materials are provided with source code for simulations, some technical details and proofs, and additional imaging results of the brain study.

Cellulose nanocrystals (CNCs), derived from cellulose, provide us with an opportunity to devise more sustainable solutions to current technological challenges. Enantioselective catalysis, especially heterogeneous, is the preferred method for the synthesis of pure chiral molecules in the fine chemical industries. Cellulose has been long sought as a chiral inducer in enantioselective catalysis. We report herein an unprecedentedly high enantiomeric excess (ee) for Pd patches deposited onto CNCs used as catalysts for the hydrogenation of prochiral ketones in water at room temperature and 4 bar H2. Our system, where CNCs acted as support and sole chiral source, achieved an ee of 65% with 100% conversions. Cryo-electron microscopy, high-resolution transmission electron microscopy, and tomography were used for the first time to study the 3D structure of a metal functionalized CNC hybrid. It established the presence of sub-nanometer-thick Pd patches at the surface of CNCs and provided insight into the chiral induction mechanism.

In order to study the improvement in reaction efficiency associated with the improved flow characteristics over static or quasi static cells, a dynamic electrochemical reactor is designed and tested. Experimental results are compared to a computational fluid dynamics (CFD) model of the cell, based only on turbulence characteristics. We intend to identify the optimal geometrical configuration, using CFD to determine how the electrodes, inlet and outlet tubes should be positionned. It is found that the overall reaction efficiency of the cell correlates almost perfectly with the volumetric turbulence intensity. The incidence angle (IA) formed by the inlet and the position of the electrodes governs the overall reaction rate.

The electrocatalytic hydrogenation (ECH) of phenol has been studied using palladium supported on gamma-alumina (10% Pd-Al2O3) catalysts. The catalyst powders were suspended in aqueous supporting electrolyte solutions containing methanol and short-chain aliphatic acids (acetic acid, propionic acid, or butyric acid) and were dynamically circulated through a reticulated vitreous carbon cathode. The efficiency of the hydrogenation process was measured as a function of the total electrolytic charge and was compared for different types of supporting electrolyte and for various solvent compositions. Our results show that these experimental parameters strongly affect the overall ECH efficiency of phenol. The ECH efficiency and yields vary inversely with the quantity of methanol present in the electrolytic solutions, whereas the presence of aliphatic carboxylic acids increased the ECH efficiency in proportion to the chain length of the specific acids employed. In all cases, ECH efficiency was directly correlated with the adsorption properties of phenol onto the Pd-alumina catalyst in the studied electrolyte solution, as measured independently using dynamic adsorption isotherms. It is shown that the alumina surface binds the aliphatic acids via the carboxylate terminations and transforms the catalyst into an organically functionalized material. Temperature-programmed mass spectrometry analysis and diffuse-reflectance infrared spectroscopy measurements confirm that the organic acids are stably bound to the alumina surface below 200 degrees C, with coverages that are independent of the acid chain length. These reproducibly functionalized alumina surfaces control the adsorption/desorption equilibrium of the target phenol molecules and allow us to prepare new electrocatalytic materials to enhance the efficiency of the ECH process. The in situ grafting of specific aliphatic acids on general purpose Pd-alumina catalysts offers a new and flexible mechanism to control the ECH process to enhance the selectivity, efficiency, and yields according to the properties of the specific target molecule.

Subdural electrocorticographic (ECoG) recordings in patients undergoing epilepsy surgery have shown that functional activation is associated with event-related broadband gamma activity in a higher frequency range (>70 Hz) than previously studied in human scalp EEG. To investigate the utility of this high gamma activity (HGA) for mapping language cortex, we compared its neuroanatomical distribution with functional maps derived from electrical cortical stimulation (ECS), which remains the gold standard for predicting functional impairment after surgery for epilepsy, tumours or vascular malformations. Thirteen patients had undergone subdural electrode implantation for the surgical management of intractable epilepsy. Subdural ECoG signals were recorded while each patient verbally named sequentially presented line drawings of objects, and estimates of event-related HGA (80-100 Hz) were made at each recording site. Routine clinical ECS mapping used a subset of the same naming stimuli at each cortical site. If ECS disrupted mouth-related motor function, i.e. if it affected the mouth, lips or tongue, naming could not be tested with ECS at the same cortical site. Because naming during ECoG involved these muscles of articulation, the sensitivity and specificity of ECoG HGA were estimated relative to both ECS-induced impairments of naming and ECS disruption of mouth-related motor function. When these estimates were made separately for 12 electrode sites per patient (the average number with significant HGA), the specificity of ECoG HGA with respect to ECS was 78% for naming and 81% for mouth-related motor function, and equivalent sensitivities were 38% and 46%, respectively. When ECS maps of naming and mouth-related motor function were combined, the specificity and sensitivity of ECoG HGA with respect to ECS were 84% and 43%, respectively. This study indicates that event-related ECoG HGA during confrontation naming predicts ECS interference with naming and mouth-related motor function with good specificity but relatively low sensitivity. Its favourable specificity suggests that ECoG HGA can be used to construct a preliminary functional map that may help identify cortical sites of lower priority for ECS mapping. Passive recordings of ECoG gamma activity may be done simultaneously at all electrode sites without the risk of after-discharges associated with ECS mapping, which must be done sequentially at pairs of electrodes. We discuss the relative merits of these two functional mapping techniques.

Penalised-spline-based additive models allow a simple mixed model representation where the variance components control departures from linear models. The smoothing parameter is the ratio of the random-coefficient and error variances and tests for linear regression reduce to tests for zero random-coefficient variances. We propose exact likelihood and restricted likelihood ratio tests for testing polynomial regression versus a general alternative modelled by penalised splines. Their spectral decompositions are used as the basis of fast simulation algorithms. We derive the asymptotic local power properties of the tests under weak conditions. In particular we characterise the local alternatives that are detected with asymptotic probability one. Confidence intervals for the smoothing parameter are obtained by inverting the tests for a fixed smoothing parameter versus a general alternative. We discuss F and R tests and show that ignoring the variability in the smoothing parameter estimator can have a dramatic effect on their null distributions. The powers of several known tests are investigated and a small set of tests with good power properties is identified. The restricted likelihood ratio test is among the best in terms of power. © 2005 Biometrika Trust.

We consider regression when the predictor is measured with error and an instrumental variable (IV) is available. The regression function can be modeled linearly, nonlinearly, or nonparametrically. Our major new result shows that the regression function and all parameters in the measurement error model are identified under relatively weak conditions, much weaker than previously known to imply identifiability. In addition, we exploit a characterization of the IV estimator as a classical “correction for attenuation” method based on a particular estimate of the variance of the measurement error. This estimate of the measurement error variance allows us to construct functional nonparametric regression estimators making no assumptions about the distribution of the unobserved predictor and structural estimators that use parametric assumptions about this distribution. The functional estimators uses simulation extrapolation or deconvolution kernels and the structural method uses Bayesian Markov chain Monte Carlo. ...

SummaryWe consider the problem of testing null hypotheses that include restrictions on the variance component in a linear mixed model with one variance component and we derive the finite sample and asymptotic distribution of the likelihood ratio test and the restricted likelihood ratio test. The spectral representations of the likelihood ratio test and the restricted likelihood ratio test statistics are used as the basis of efficient simulation algorithms of their null distributions. The large sample χ2 mixture approximations using the usual asymptotic theory for a null hypothesis on the boundary of the parameter space have been shown to be poor in simulation studies. Our asymptotic calculations explain these empirical results. The theory of Self and Liang applies only to linear mixed models for which the data vector can be partitioned into a large number of independent and identically distributed subvectors. One-way analysis of variance and penalized splines models illustrate the results.

The ELISA tests that are available to detect an infection with Mycobacterium avium subsp. paratuberculosis (MAP) have a limited validity expressed as the sensitivity (Se) and specificity (Sp). In many studies, the Se and Sp of the tests are treated as constants and this will result in an underestimation of the variability of the true prevalence (TP). Bayesian inference provided a natural framework for using information on the test variability (i.e., the uncertainty) in the estimates of test Se and Sp when estimating the TP. Data from two prevalence studies for MAP using an ELISA in several regions in two locations were available for the analyses. In location 1, all cattle of at least 3 years of age were sampled in approximately 90 randomly sampled herds in each of the four regions of the country. In location 2, in 30 randomly sampled herds in each of three regions, approximately 30 randomly selected cows were sampled. Information about the unknown test Se and Sp and MAP prevalence was incorporated into a Bayesian model by joint prior probability distributions. Posterior estimates were obtained by combining the actual likelihood with the prior distributions using Bayes' formula. The corrected cow-level TP (proportion of infected cows in a herd) was low, 5.8 and 3.6% in locations 1 and 2, respectively. Certain regions within a location differed significantly in herd-level TP (proportion of infected herds). The herd-level TP was 54.3% in location 1 (95% credible interval (CI) 46.1, 63.3%) and 32.9% in location 2 (95% CI: 14.4, 73.3%). The variation in the herd-level TP estimate for location 2 was more than three times as large as the variation in location 1 mainly because of the relatively small number of investigated herds in location 2. In future prevalence studies for MAP, sample size calculations should be based on a very low cow-level prevalence. Approximately 50 and 90% of the herds in the current study had an estimated cow-level TP below 4 and 10%, respectively.

We develop a flexible framework for modeling high-dimensional imaging data observed longitudinally. The approach decomposes the observed variability of repeatedly measured high-dimensional observations into three additive components: a subject-specific imaging random intercept that quantifies the cross-sectional variability, a subject-specific imaging slope that quantifies the dynamic irreversible deformation over multiple realizations, and a subject-visit-specific imaging deviation that quantifies exchangeable effects between visits. The proposed method is very fast, scalable to studies including ultrahigh-dimensional data, and can easily be adapted to and executed on modest computing infrastructures. The method is applied to the longitudinal analysis of diffusion tensor imaging (DTI) data of the corpus callosum of multiple sclerosis (MS) subjects. The study includes $176$ subjects observed at $466$ visits. For each subject and visit the study contains a registered DTI scan of the corpus callosum at roughly 30,000 voxels., Comment: Published in at http://dx.doi.org/10.1214/14-AOAS748 the Annals of Applied Statistics (http://www.imstat.org/aoas/) by the Institute of Mathematical Statistics (http://www.imstat.org)

This is a copy of the author 's final draft version of an article published in the journal Transactions on computational collective intelligence. The final publication is available at Springer via http://dx.doi.org/10.1007/978-3-662-49017-4_6 Vehicular Data includes different facts and measurements made over a set of moving vehicles. Most of us use cars or public transportation for our work commute, daily routines and leisure. But, except of our destination, possible time of arrival and what is directly around us, we know very little about the traffic conditions in the city as a whole. Because all roads are connected in a vast network, events in other parts of town can and will directly affect us. The more we know about the traffic inside a city, the better decisions we can make. Vehicular measurements may contain a vast amount of information about the way our cities function. Information that can be used for more than improving our commute, it is indicative of other features of the city like the amount of pollution in different regions. All the information and knowledge we can extract, can be used to directly improve our life. We live in a world where data is constantly generated and we store it and process it at an ever growing rate. Vehicular Data does not stray from this fact and is rapidly growing in size and complexity, with more and more ways to monitoring traffic, either from inside cars or from sensors placed on the road. Smartphones and in-car-computers are now common and they can produce a vast amount of data: it can identify a cars location, destination, current speed and even driving habits. Machine learning is the perfect complement for Big Data, as large data sets can be rendered useless without methods to extract knowledge and information from them. Machine learning, currently a popular research topic, has a large number of algorithms design to achieve this task, of knowledge extraction. Most of these techniques and algorithms can be directly applied to Vehicular Data. In this article we demonstrate how the use of a simple algorithm, k-Nearest Neighbors, can be used to extract valuable information from even a relatively small vehicular data set. Because of the vast size of most of our cities and the number of cars that are on their roads at any time of the day, standard machine learning systems do not manage to process data in a manner that would permit real time use of the extracted information. A solution to this problem is brought by distributed systems and cloud processing. By parallelizing and distributing machine learning algorithms we can use data at its highest potential and with little delay. Here, we show how this can be achieved by distributing the k-Nearest Neighbors machine learning algorithm over MPI. We hope this would motivate the research into other combinations of merging machine learning algorithms with Vehicular Data sets.

Structural magnetic resonance imaging (MRI) can be used to detect lesions in the brains of multiple sclerosis (MS) patients. The formation of these lesions is a complex process involving inflammation, tissue damage, and tissue repair, all of which are visible on MRI. Here we characterize the lesion formation process on longitudinal, multi-sequence structural MRI from 34 MS patients and relate the longitudinal changes we observe within lesions to therapeutic interventions. In this article, we first outline a pipeline to extract voxel level, multi-sequence longitudinal profiles from four MRI sequences within lesion tissue. We then propose two models to relate clinical covariates to the longitudinal profiles. The first model is a principal component analysis (PCA) regression model, which collapses the information from all four profiles into a scalar value. We find that the score on the first PC identifies areas of slow, long-term intensity changes within the lesion at a voxel level, as validated by two experienced clinicians, a neuroradiologist and a neurologist. On a quality scale of 1 to 4 (4 being the highest) the neuroradiologist gave the score on the first PC a median rating of 4 (95% CI: [4,4]), and the neurologist gave it a median rating of 3 (95% CI: [3,3]). In the PCA regression model, we find that treatment with disease modifying therapies (p-value < 0.01), steroids (p-value < 0.01), and being closer to the boundary of abnormal signal intensity (p-value < 0.01) are associated with a return of a voxel to intensity values closer to that of normal-appearing tissue. The second model is a function-on-scalar regression, which allows for assessment of the individual time points at which the covariates are associated with the profiles. In the function-on-scalar regression both age and distance to the boundary were found to have a statistically significant association with the profiles.

We examine whether the distribution of physical activity during the day acts as an independent predictor of mortality in National Health and Nutrition Examination Survey 2003–2006. Time of day analyses were conducted using 12 time windows of 2 hours represented by the percentage of the total log-transformed activity count (pTLAC) spent in each. In models adjusted for age, sex, race/ethnicity, education, body mass index, comorbidities, and the total log-transformed activity count, a higher percentage of total physical activity between 10am-12pm (HR: 0.96, p-value = 0.003) and a lower percentage of total physical activity between 8pm-4am (HR: 1.03, p-value < 0.01) were significantly associated with lower risk of mortality. These findings help define patterns of physical activity that are beneficial to health, and hold significant implications for future interventions aimed at increasing physical activity and enhancing sleep quality.

Accelerometers have been widely deployed in public health studies in recent years and research has mainly focused on summarized metrics provided by accelerometers manufactures, such as the activity counts (AC). Such measures do not have a publicly available formula and can vary by device manufacturer. To address these problems, we developed the activity index (AI), a new metric for summarizing raw tri-axial accelerometry data, and compared the AI to AC’s performance for distinguishing various types of activities and estimating energy expenditure. The analysis was conducted using data from the Women’s Health Initiative, in which tri-axial raw acceleration data and energy expenditure were collected at the same time. ROC analyses indicated that AI better distinguished between different types of activities than AC. AI better associated with METs as well. The proposed AI provides a transparent and reliable way to summarize densely sampled raw acceleration data.

Commonly used measures of physical activity (PA) fail to account for heterogeneity in maximal capacity between individuals. Utilizing the concept of heart rate reserve, we estimated minute level physical exertion (PE) among BLSA subjects with at least 3 days of Actiheart data (n = 411). Subjects’ average time spent in each of sedentary, light, moderate and vigorous PE was associated with age adjusting for demographics, chronic conditions, and PA. Despite a significant decline in total PA with age (p < 0.001), time spent in moderate and vigorous PE is positively associated with age in both women (pmoderate < 0.001, pvigorous < 0.001) and men (pmoderate = 0.137, pvigorous = 0.222). Chronic conditions are also generally associated with increased average PE. Our results suggest that individuals with decreased capacity are, on average, engaging moderate and high relative intensity activities at a rate comparable to, or higher than, that of healthier individuals.

X-ray computed tomography (CT) imaging of the brain is commonly used in diagnostic settings. Although CT scans are primarily used in clinical practice, they are increasingly used in research. A fundamental processing step in brain imaging research is brain extraction - the process of separating the brain tissue from all other tissues. Methods for brain extraction have either been 1) validated but not fully automated, or 2) fully automated and informally proposed, but never formally validated.To systematically analyze and validate the performance of FSL's brain extraction tool (BET) on head CT images of patients with intracranial hemorrhage. This was done by comparing the manual gold standard with the results of several versions of automatic brain extraction and by estimating the reliability of automated segmentation of longitudinal scans. The effects of the choice of BET parameters and data smoothing is studied and reported.All images were thresholded using a 0-100 Hounsfield unit (HU) range. In one variant of the pipeline, data were smoothed using a 3-dimensional Gaussian kernel (σ=1mm(3)) and re-thresholded to 0-100HU; in the other, data were not smoothed. BET was applied using 1 of 3 fractional intensity (FI) thresholds: 0.01, 0.1, or 0.35 and any holes in the brain mask were filled. For validation against a manual segmentation, 36 images from patients with intracranial hemorrhage were selected from 19 different centers from the MISTIE (Minimally Invasive Surgery plus recombinant-tissue plasminogen activator for Intracerebral Evacuation) stroke trial. Intracranial masks of the brain were manually created by one expert CT reader. The resulting brain tissue masks were quantitatively compared to the manual segmentations using sensitivity, specificity, accuracy, and the Dice Similarity Index (DSI). Brain extraction performance across smoothing and FI thresholds was compared using the Wilcoxon signed-rank test. The intracranial volume (ICV) of each scan was estimated by multiplying the number of voxels in the brain mask by the dimensions of each voxel for that scan. From this, we calculated the ICV ratio comparing manual and automated segmentation: ICVautomated/ICVmanual. To estimate the performance in a large number of scans, brain masks were generated from the 6 BET pipelines for 1095 longitudinal scans from 129 patients. Failure rates were estimated from visual inspection. ICV of each scan was estimated and an intraclass correlation (ICC) was estimated using a one-way ANOVA.Smoothing images improves brain extraction results using BET for all measures except specificity (all p0.01, uncorrected), irrespective of the FI threshold. Using an FI of 0.01 or 0.1 performed better than 0.35. Thus, all reported results refer only to smoothed data using an FI of 0.01 or 0.1. Using an FI of 0.01 had a higher median sensitivity (0.9901) than an FI of 0.1 (0.9884, median difference: 0.0014, p0.001), accuracy (0.9971 vs. 0.9971; median difference: 0.0001, p0.001), and DSI (0.9895 vs. 0.9894; median difference: 0.0004, p0.001) and lower specificity (0.9981 vs. 0.9982; median difference: -0.0001, p0.001). These measures are all very high indicating that a range of FI values may produce visually indistinguishable brain extractions. Using smoothed data and an FI of 0.01, the mean (SD) ICV ratio was 1.002 (0.008); the mean being close to 1 indicates the ICV estimates are similar for automated and manual segmentation. In the 1095 longitudinal scans, this pipeline had a low failure rate (5.2%) and the ICC estimate was high (0.929, 95% CI: 0.91, 0.945) for successfully extracted brains.BET performs well at brain extraction on thresholded, 1mm(3) smoothed CT images with an FI of 0.01 or 0.1. Smoothing before applying BET is an important step not previously discussed in the literature. Analysis code is provided.

Motivated by modern observational studies, we introduce a class of functional models that expands nested and crossed designs. These models account for the natural inheritance of correlation structure from sampling design in studies where the fundamental sampling unit is a function or image. Inference is based on functional quadratics and their relationship with the underlying covariance structure of the latent processes. A computationally fast and scalable estimation procedure is developed for ultra-high dimensional data. Methods are illustrated in three examples: high-frequency accelerometer data for daily activity, pitch linguistic data for phonetic analysis, and EEG data for studying electrical brain activity during sleep.