Association between Perfluoroalkyl and Polyfluoroalkyl Substances and Women's Infertility, NHANES 2013-2016

Perfluoroalkyl and polyfluoroalkyl substances (PFASs) are widely used in consumer products. However, the role of PFAS in infertility is still poorly understood. A total of 788 women from the 2013-2016 nationally representative NHANES were included to explore the association between PFAS exposure and self-reported infertility. Six PFAS, including PFDE, PFNA, PFHxS, n-PFOA, n-PFOS, and Sm-PFOS, were detected by online SPE-HPLC-TIS-MS/MS. We used the generalized linear regression model (GLM), generalized additive models (GAM), and Bayesian kernel machine regression (BKMR) to assess the single effects, non-linear relationships, and mixed effects on women's infertility, respectively. The prevalence of self-reported infertility was 15.54% in this study. In GLM, n-PFOA showed a negative association with self-reported infertility in women for the Q3 (OR: 0.396, 95% CI: 0.119, 0.788) and Q4 (OR: 0.380, 95% CI: 0.172-0.842) compared with Q1 (p for trend = 0.013). A negative trend was also observed in n-PFOS and ∑PFOS (p for trend < 0.05). In GAM, a non-linear relationship was revealed in Sm-PFOS, which exhibits a U-shaped relationship. The BKMR model indicated that there might be a joint effect between PFAS and women's infertility, to which PFNA contributed the highest effect (PIP = 0.435). Moreover, age stratification analysis showed a different dose-response curve in under and above 35 years old. Women under the age of 35 have a more noticeable U-shaped relationship with infertility. Therefore, the relatively low level of mixed PFAS exposure was negatively associated with self-reported infertility in women in general, and the impact of PFAS on infertility may vary among women of different age groups. Further studies are needed to determine the etiological relationship.

Generalized models for quantifying laterality using functional transcranial Doppler ultrasound

We consider how analysis of brain lateralization using functional transcranial Doppler ultrasound (fTCD) data can be brought in line with modern statistical methods typically used in functional magnetic resonance imaging (fMRI). Conventionally, a laterality index is computed in fTCD from the difference between the averages of each hemisphere's signal within a period of interest (POI) over a series of trials. We demonstrate use of generalized linear models (GLMs) and generalized additive models (GAM) to analyze data from individual participants in three published studies (N = 154, 73 and 31), and compare this with results from the conventional POI averaging approach, and with laterality assessed using fMRI (N = 31). The GLM approach was based on classic fMRI analysis that includes a hemodynamic response function as a predictor; the GAM approach estimated the response function from the data, including a term for time relative to epoch start (simple GAM), plus a categorical index corresponding to individual epochs (complex GAM). Individual estimates of the fTCD laterality index are similar across all methods, but error of measurement is lowest using complex GAM. Reliable identification of cases of bilateral language appears to be more accurate with complex GAM. We also show that the GAM-based approach can be used to efficiently analyze more complex designs that incorporate interactions between tasks.

Calculating and Comparing the Annualized Relapse Rate and Estimating the Confidence Interval in Relapsing Neurological Diseases

Calculating the crude or adjusted annualized relapse rate (ARR) and its confidence interval (CI) is often required in clinical studies to evaluate chronic relapsing diseases, such as multiple sclerosis and neuromyelitis optica spectrum disorders. However, accurately calculating ARR and estimating the 95% CI requires careful application of statistical approaches and basic familiarity with the exponential family of distributions. When the relapse rate can be regarded as constant over time or by individuals, the crude ARR can be calculated using the person-years method, which divides the number of all observed relapses among all participants by the total follow-up period of the study cohort. If the number of relapses can be modeled by the Poisson distribution, the 95% CI of ARR can be obtained by finding the 2.5% upper and lower critical values of the parameter λ as the mean. Basic familiarity with F-statistics is also required when comparing the ARR between two disease groups. It is necessary to distinguish the observed relapse rate ratio (RR) between two sample groups (sample RR) from the unobserved RR between their originating populations (population RR). The ratio of population RR to sample RR roughly follows the F distribution, with degrees of freedom obtained by doubling the number of observed relapses in the two sample groups. Based on this, a 95% CI of the population RR can be estimated. When the count data of the response variable is overdispersed, the negative binomial distribution would be a better fit than the Poisson. Adjusted ARR and the 95% CI can be obtained by using the generalized linear regression models after selecting appropriate error structures (e.g., Poisson, negative binomial, zero-inflated Poisson, and zero-inflated negative binomial) according to the overdispersion and zero-inflation in the response variable.

Modeling dragonfly population data with a Bayesian bivariate geometric mixed-effects model

We develop a generalized linear mixed model (GLMM) for bivariate count responses for statistically analyzing dragonfly population data from the Northern Netherlands. The populations of the threatened dragonfly species Aeshna viridis were counted in the years 2015-2018 at 17 different locations (ponds and ditches). Two different widely applied population size measures were used to quantify the population sizes, namely the number of found exoskeletons ('exuviae') and the number of spotted egg-laying females were counted. Since both measures (responses) led to many zero counts but also feature very large counts, our GLMM model builds on a zero-inflated bivariate geometric (ZIBGe) distribution, for which we show that it can be easily parameterized in terms of a correlation parameter and its two marginal medians. We model the medians with linear combinations of fixed (environmental covariates) and random (location-specific intercepts) effects. Modeling the medians yields a decreased sensitivity to overly large counts; in particular, in light of growing marginal zero inflation rates. Because of the relatively small sample size (n = 114) we follow a Bayesian modeling approach and use Metropolis-Hastings Markov Chain Monte Carlo (MCMC) simulations for generating posterior samples.

Localizing Epileptic Foci Using Simultaneous EEG-fMRI Recording: Template Component Cross-Correlation

Conventional EEG-fMRI methods have been proven to be of limited use in the sense that they cannot reveal the information existing in between the spikes. To resolve this issue, the current study obtains the epileptic components time series detected on EEG and uses them to fit the Generalized Linear Model (GLM), as a substitution for classical regressors. This approach allows for a more precise localization, and equally importantly, the prediction of the future behavior of the epileptic generators. The proposed method approaches the localization process in the component domain, rather than the electrode domain (EEG), and localizes the generators through investigating the spatial correlation between the candidate components and the spike template, as well as the medical records of the patient. To evaluate the contribution of EEG-fMRI and concordance between fMRI and EEG, this method was applied on the data of 30 patients with refractory epilepsy. The results demonstrated the significant numbers of 29 and 24 for concordance and contribution, respectively, which mark improvement as compared to the existing literature. This study also shows that while conventional methods often fail to properly localize the epileptogenic zones in deep brain structures, the proposed method can be of particular use. For further evaluation, the concordance level between IED-related BOLD clusters and Seizure Onset Zone (SOZ) has been quantitatively investigated by measuring the distance between IED/SOZ locations and the BOLD clusters in all patients. The results showed the superiority of the proposed method in delineating the spike-generating network compared to conventional EEG-fMRI approaches. In all, the proposed method goes beyond the conventional methods by breaking the dependency on spikes and using the outside-the-scanner spike templates and the selected components, achieving an accuracy of 97%. Doing so, this method contributes to improving the yield of EEG-fMRI and creates a more realistic perception of the neural behavior of epileptic generators which is almost without precedent in the literature.

Technical Note: On the development of an outcome-driven frequency filter for improving Radiomics-based modeling of Human Papilloma Virus (HPV) in patients with oropharyngeal squamous cell carcinomas

PURPOSE

To implement an outcome-driven frequency filter for improving radiomics-based modeling of human papilloma virus (HPV) for patients with oropharyngeal squamous cell carcinoma (OPSCC).

METHODS AND MATERIALS

One hundred twenty-eight OPSCC patients with known HPV status (60-HPV+ and 68-HPV-, confirmed by immunohistochemistry-P16 protein testing) were retrospectively studied. A 3D Discrete Fourier Transform was applied on contrast-enhanced CT images of patient gross tumor volumes (GTV's) to transform intensity distributions to the frequency domain and estimate frequency power spectrums of HPV- and HPV+ patient cohorts. Statistical analyses were performed to rank frequency bands contributing towards prediction of HPV status. An outcome-driven frequency filter was designed accordingly and applied to GTV frequency information. 3D Inverse-Discrete-Fourier-Transform was applied to reconstruct HPV-related frequency-filtered images. Radiomics features (11 feature-categories) were extracted from pre- and post- frequency filtered images using our previously published 'ROdiomiX' software. Least-Absolute-Shrinkage-and-Selection-Operation (Lasso) combined with a Generalized-Linear-Model (Lasso-GLM) was developed to identify and rank feature subsets with optimal information for prediction of HPV+/-. Radiomics-based Lasso-GLM classifiers (pre- and post-frequency filtered) were constructed and validated using a random permutation sampling and nested cross-validation techniques. Average Area Under Receiver Operating Characteristic (AUC), and Positive and Negative Predictive values (PPV, NPV) were computed to estimate generalization error and prediction performance.

RESULTS

Among 192 radiomic features, 15 features were found to be statistically significant discriminators between HPV+/- cohorts on post-frequency filtered CE-CT images; 14 such radiomic features were observed on pre-frequency filtered datasets. Discriminant features included tumor morphology and intensity contrast. Performances for prediction of HPV for the pre- and post-frequency filtered Lasso-GLM classifiers were: AUC/PPV/NPV = 0.789/0.755/0.805 and 0.850/0.808/0.877 respectively. Nested-CV performances for prediction of HPV for the pre- and post-frequency filtered Lasso-GLM classifiers were: AUC/PPV/NPV = 0.814/0.725/0.877 and 0.890/0.820/0.911 respectively.

CONCLUSION

Albeit subject to confirmation in a larger cohort, this pilot study presents encouraging results on the importance of frequency analysis prior to radiomic feature extraction toward enhancement of model performance for characterizing HPV in patients with OPSCC. This article is protected by copyright. All rights reserved.

A Longitudinal Analysis of Gambling Predictors among Adolescents

Although gambling is forbidden for minors, the prevalence of gambling among adolescents is increasing. In order to improve preventive interventions, more evidence on predictors of gambling onset is needed. A longitudinal study was proposed to (1) establish the prevalence of gambling; (2) identify factors associated with gambling behavior the following year; and (3) adjust a model to predict gambling behavior. A cohort of 1074 students (13-18 years old) was followed for 12 months. The prevalence of gambling reached 42.0% in the second measure. Boys gambled 2.7 times more than girls, and the highest percentages of gambling onset showed up between 13 and 14 years old. Gambling onset and maintenance was associated with gender, age, sensation-seeking, risk perception, self-efficacy for not gambling, parents' attitude towards gambling, group pressure (friends), subjective norm, exposure to advertising, accessibility, normative perception, gambling in T₁ and parents gambling behavior. Gender, gambling in T₁ and risk perception were significant in all three logistic adjusted regression models, with the fourth variable being sensation seeking, peer pressure (friends) and accessibility, respectively. It is suggested that universal prevention should be aimed preferably at children under 15 years old and to alert regulators and public administrations to the directly proportional relationship between accessibility and gambling onset.

Capturing Multiple Timescales of Adaptation to Second-Order Statistics With Generalized Linear Models: Gain Scaling and Fractional Differentiation

Single neurons can dynamically change the gain of their spiking responses to take into account shifts in stimulus variance. Moreover, gain adaptation can occur across multiple timescales. Here, we examine the ability of a simple statistical model of spike trains, the generalized linear model (GLM), to account for these adaptive effects. The GLM describes spiking as a Poisson process whose rate depends on a linear combination of the stimulus and recent spike history. The GLM successfully replicates gain scaling observed in Hodgkin-Huxley simulations of cortical neurons that occurs when the ratio of spike-generating potassium and sodium conductances approaches one. Gain scaling in the GLM depends on the length and shape of the spike history filter. Additionally, the GLM captures adaptation that occurs over multiple timescales as a fractional derivative of the stimulus envelope, which has been observed in neurons that include long timescale afterhyperpolarization conductances. Fractional differentiation in GLMs requires long spike history that span several seconds. Together, these results demonstrate that the GLM provides a tractable statistical approach for examining single-neuron adaptive computations in response to changes in stimulus variance.

Non-motor Brain Regions in Non-dominant Hemisphere Are Influential in Decoding Movement Speed

Sensorimotor control studies have predominantly focused on how motor regions of the brain relay basic movement-related information such as position and velocity. However, motor control is often complex, involving the integration of sensory information, planning, visuomotor tracking, spatial mapping, retrieval and storage of memories, and may even be emotionally driven. This suggests that many more regions in the brain are involved beyond premotor and motor cortices. In this study, we exploited an experimental setup wherein activity from over 87 non-motor structures of the brain were recorded in eight human subjects executing a center-out motor task. The subjects were implanted with depth electrodes for clinical purposes. Using training data, we constructed subject-specific models that related spectral power of neural activity in six different frequency bands as well as a combined model containing the aggregation of multiple frequency bands to movement speed. We then tested the models by evaluating their ability to decode movement speed from neural activity in the test data set. The best models achieved a correlation of 0.38 ± 0.03 (mean ± standard deviation). Further, the decoded speeds matched the categorical representation of the test trials as correct or incorrect with an accuracy of 70 ± 2.75% across subjects. These models included features from regions such as the right hippocampus, left and right middle temporal gyrus, intraparietal sulcus, and left fusiform gyrus across multiple frequency bands. Perhaps more interestingly, we observed that the non-dominant hemisphere (ipsilateral to dominant hand) was most influential in decoding movement speed.

Estimation of Symptom Severity Scores for Patients with Schizophrenia Using ERP Source Activations during a Facial Affect Discrimination Task

Precise diagnosis of psychiatric diseases and a comprehensive assessment of a patient's symptom severity are important in order to establish a successful treatment strategy for each patient. Although great efforts have been devoted to searching for diagnostic biomarkers of schizophrenia over the past several decades, no study has yet investigated how accurately these biomarkers are able to estimate an individual patient's symptom severity. In this study, we applied electrophysiological biomarkers obtained from electroencephalography (EEG) analyses to an estimation of symptom severity scores of patients with schizophrenia. EEG signals were recorded from 23 patients while they performed a facial affect discrimination task. Based on the source current density analysis results, we extracted voxels that showed a strong correlation between source activity and symptom scores. We then built a prediction model to estimate the symptom severity scores of each patient using the source activations of the selected voxels. The symptom scores of the Positive and Negative Syndrome Scale (PANSS) were estimated using the linear prediction model. The results of leave-one-out cross validation (LOOCV) showed that the mean errors of the estimated symptom scores were 3.34 ± 2.40 and 3.90 ± 3.01 for the Positive and Negative PANSS scores, respectively. The current pilot study is the first attempt to estimate symptom severity scores in schizophrenia using quantitative EEG features. It is expected that the present method can be extended to other cognitive paradigms or other psychological illnesses.

Enhancing the Biological Relevance of Secretome-Based Proteomics by Linking Tumor Cell Proliferation and Protein Secretion

Secretome profiling has become a methodology of choice for the identification of tumor biomarkers. We hypothesized that due to the dynamic nature of secretomes cellular perturbations could affect their composition but also change the global amount of protein secreted per cell. We confirmed our hypothesis by measuring the levels of secreted proteins taking into account the amount of proteome produced per cell. Then, we established a correlation between cell proliferation and protein secretion that explained the observed changes in global protein secretion. Next, we implemented a normalization correcting the statistical results of secretome studies by the global protein secretion of cells into a generalized linear model (GLM). The application of the normalization to two biological perturbations on tumor cells resulted in drastic changes in the list of statistically significant proteins. Furthermore, we found that known epithelial-to-mesenchymal transition (EMT) effectors were only statistically significant when the normalization was applied. Therefore, the normalization proposed here increases the sensitivity of statistical tests by increasing the number of true-positives. From an oncology perspective, the correlation between protein secretion and cellular proliferation suggests that slow-growing tumors could have high-protein secretion rates and consequently contribute strongly to tumor paracrine signaling.

Tensor Regression with Applications in Neuroimaging Data Analysis

Classical regression methods treat covariates as a vector and estimate a corresponding vector of regression coefficients. Modern applications in medical imaging generate covariates of more complex form such as multidimensional arrays (tensors). Traditional statistical and computational methods are proving insufficient for analysis of these high-throughput data due to their ultrahigh dimensionality as well as complex structure. In this article, we propose a new family of tensor regression models that efficiently exploit the special structure of tensor covariates. Under this framework, ultrahigh dimensionality is reduced to a manageable level, resulting in efficient estimation and prediction. A fast and highly scalable estimation algorithm is proposed for maximum likelihood estimation and its associated asymptotic properties are studied. Effectiveness of the new methods is demonstrated on both synthetic and real MRI imaging data.