Assessing Methods for Evaluating the Number of Components in Non-Negative Matrix Factorization.

Non-negative matrix factorization is a relatively new method of matrix decomposition which factors an m×n data matrix X into an m×k matrix W and a k×n matrix H, so that X≈W×H. Importantly, all values in X, W, and H are constrained to be non-negative. NMF can be used for dimensionality reduction, since the k columns of W can be considered components into which X has been decomposed. The question arises: how does one choose k? In this paper, we first assess methods for estimating k in the context of NMF in synthetic data. Second, we examine the effect of normalization on this estimate's accuracy in empirical data. In synthetic data with orthogonal underlying components, methods based on PCA and Brunet's Cophenetic Correlation Coefficient achieved the highest accuracy. When evaluated on a well-known real dataset, normalization had an unpredictable effect on the estimate. For any given normalization method, the methods for estimating k gave widely varying results. We conclude that when estimating k, it is best not to apply normalization. If underlying components are known to be orthogonal, then Velicer's MAP or Minka's Laplace-PCA method might be best. However, when orthogonality of the underlying components is unknown, none of the methods seemed preferable.

Urinary paracetamol and time-to-pregnancy.

STUDY QUESTION: Is preconception urinary paracetamol (acetaminophen) associated with time-to-pregnancy (TTP)? SUMMARY ANSWER: Higher urinary paracetamol concentrations among male partners were associated with a longer TTP. WHAT IS KNOWN ALREADY: Paracetamol is a commonly used analgesic among women and men of all ages. As metabolites of select chemicals used in the manufacturing of polyurethane foam, dyes and various industrial products, as well as a common medicinal product, paracetamol and its primary metabolite p-aminophenol, are ubiquitous in the environment. Studies investigating the relationship between adult urinary concentrations of paracetamol and TTP are lacking. STUDY DESIGN, SIZE, DURATION: This prospective cohort included 501 couples discontinuing contraception for the purposes of attempting conception during the years 2005-2009 and residing in Michigan or Texas, USA. PARTICIPANTS/MATERIALS, SETTING, METHODS: Total urinary paracetamol, its metabolite para-aminophenol (p-aminophenol), and a summary measure of both urinary biomarkers were quantified by ultra-performance liquid chromatography coupled with an electrospray triple quadrupole mass spectrometry (UPLC-ESI-MS/MS). Female partners used the Clearblue® digital home test to confirm pregnancy. Cox's proportional odds models for discrete survival time were used to estimate fecundability odds ratios (FORs) and 95% confidence intervals (CIs), adjusting for age, body mass index (BMI), urinary creatinine, preconception smoking status, race/ethnicity and household income. Models were further adjusted for hypothyroidism and hypertension as an attempt to account for possible indications of paracetamol medication use. FOR estimates <1.0 denote a longer TTP (diminished fecundity). Models were performed to examine urinary concentrations of paracetamol as a continuous and variable or categorized into quartiles. In light of TTP being a couple-dependent outcome, models were first performed for females and males, modeled separately, and then modeled for couples with each partner's concentrations being adjusted for the other. MAIN RESULTS AND THE ROLE OF CHANCE: Among the 501 enrolled couples, 347 (69%) had an human chorionic gonadotrophin confirmed pregnancy. Urinary concentrations of paracetamol were lowest among females and males who achieved pregnancy and p-aminophenol concentrations were lowest among those not achieving pregnancy. Urinary paracetamol concentrations were higher among female than male partners (Median = 26.6 and 13.2 ng/ml, respectively; P < 0.0001). After adjustment for age, BMI, urinary creatinine, preconception smoking status, race/ethnicity and household income, the highest quartile of male urinary paracetamol was associated with a longer TTP [FOR = 0.67; 95% CI = (0.47, 0.95)]. This association remained after adjustment for chronic health conditions, hypothyroidism and hypertension and female partner's urinary paracetamol concentration [FOR = 0.65; 95% CI = (0.45, 0.94)]. No associations were observed between female or male partners' urinary concentrations of paracetamol or of its metabolite p-aminophenol when urinary concentrations were modeled continuously. LIMITATIONS, REASONS FOR CAUTION: Only a single spot urine was available for analysis despite the short-lived nature of paracetamol. Additionally, participants were not asked to provide information on indication of use for paracetamol medications; any underlying conditions for the paracetamol use would have been potential confounders. WIDER IMPLICATIONS OF THE FINDINGS: If corroborated with more robust studies, findings from our exploratory analysis may have both clinical and public health relevance among reproductive aged individuals, including those trying for pregnancy, given the prevalent use of paracetamol/acetaminophen medications and the ubiquitous nature of paracetamol in the environment. STUDY FUNDING/COMPETING INTERESTS: This research was supported by the National Institutes of Health, Intramural Research Program, and Eunice Kennedy Shriver National Institute of Child Health and Human Development (contracts N01-HD-3-3355; N01-HD-3-3356; NOH-HD-3-3358; HHSN27500001/HHSN27500001). None of the authors have any conflicts to declare.

Time-to-Pregnancy Associated With Couples' Use of Tobacco Products.

INTRODUCTION: Previous studies suggest female smoking increases time-to-pregnancy (TTP), a couple-dependent reproductive outcome, while associations with male smoking are more ambiguous. Furthermore, despite small increases in smokeless tobacco use in the United States, no prior study has evaluated TTP among smokeless tobacco users. METHODS: Using population-based sampling in 16 counties in Michigan and Texas, 501 couples discontinuing contraception to become pregnant were followed until positive pregnancy test or 12 months of trying. Participants were interviewed on lifetime and current cigarette, cigar, and chew/snuff (smokeless) use and provided blood samples for quantification of heavy metals and cotinine. Fecundability odds ratios (FORs) and 95% confidence intervals (95% CIs) were estimated, adjusted for demographics/lifestyle. FORs less than 1 reflect longer TTP. RESULTS: Eleven percentage of females and 15% of males smoked cigarettes. Among men, 14% smoked cigars, 9% used snuff, and 2% used chew. Compared with never tobacco users, male (FOR: 0.41, 95% CI: 0.24, 0.68) and female (FOR: 0.53, 95% CI: 0.33, 0.85) smoking were individually associated with longer TTP; males' smoking remained significant (FOR: 0.46, 95% CI: 0.27, 0.79) when modeling partners together. Cadmium levels were higher in smokers than smokeless tobacco and never users; adjusting for cadmium attenuated the cigarette-TTP association, particularly among women. TTP was shorter among smokeless tobacco users relative to smokers (FOR: 2.86, 95% CI: 1.47, 5.57). CONCLUSIONS: Compared with never users, smokeless tobacco did not alter TTP in our cohort; however, TTP was shorter compared with smokers. We observed longer TTP in male and female smokers; cadmium may partially contribute. IMPLICATIONS: Both partners' preconception smoking contributed to longer TTP, highlighting the importance of both partners' lifestyles in healthy reproduction and underscores the need for couple-based preconception guidance. The male's contribution is a new finding. Higher cadmium levels may partially contribute to longer TTP in smokers, particularly among females. Though we do not observe longer TTP among a small sample of smokeless tobacco users compared with never tobacco users, we observe shorter TTP compared with smokers. Further work is needed to more thoroughly delineate the relationship between smokeless tobacco use and TTP and possible mechanisms of tobacco use's effects on reproduction.

Lifestyle and pregnancy loss in a contemporary cohort of women recruited before conception: The LIFE Study.

OBJECTIVE: To estimate pregnancy loss incidence in a contemporary cohort of couples whose lifestyles were measured during sensitive windows of reproduction to identify factors associated with pregnancy loss for the continual refinement of preconception guidance. DESIGN: Prospective cohort with preconception enrollment. SETTING: Sixteen counties in Michigan and Texas. PATIENT(S): Three hundred forty-four couples with a singleton pregnancy followed daily through 7 postconception weeks of gestation. INTERVENTION(S): None. Couples daily recorded use of cigarettes, caffeinated and alcoholic beverages, and multivitamins. Women used fertility monitors for ovulation detection and digital pregnancy tests. Pregnancy loss was denoted by conversion to a negative pregnancy test, onset of menses, or clinical confirmation depending upon gestation. Using proportional hazards regression and accounting for right censoring, we estimated adjusted hazard ratios and 95% confidence intervals (aHR, 95% CI) for couples' lifestyles (cigarette smoking, alcoholic and caffeinated drinks, multivitamins) during three sensitive windows: preconception, early pregnancy, and periconception. MAIN OUTCOME MEASURE(S): Incidence and risk factors for pregnancy loss. RESULT(S): Ninety-eight of 344 (28%) women with a singleton pregnancy experienced an observed pregnancy loss. In the preconception window, loss was associated with female age ≥35 years (1.96, 1.13-3.38) accounting for couples' ages, women's and men's consumption of >2 daily caffeinated beverages (1.74, 1.07-2.81; and 1.73, 1.10-2.72, respectively), and women's vitamin adherence (0.45, 0.25-0.80). The findings were similar for lifestyle during the early pregnancy and periconception windows. CONCLUSION(S): Couples' preconception lifestyle factors were associated with pregnancy loss, although women's multivitamin adherence dramatically reduced risk. The findings support continual refinement and implementation of preconception guidance.

Persistent organic pollutants and pregnancy complications.

We sought to investigate the relationship between maternal preconception exposures to persistent organic pollutants (POPs) and pregnancy complications, gestational diabetes (GDM) and gestational hypertension. Data from 258 (51%) women with human chorionic gonadotropin (hCG) confirmed pregnancies reaching ≥24weeks gestation, from a prospective cohort of 501 couples who discontinued contraception to attempt pregnancy, were analyzed. Preconception concentrations of 9 organochlorine pesticides (OCPs) and 10 polybrominated diphenyl ethers (PBDEs) were quantified in serum. In separate multiple logistic regression models of self-reported physician diagnosed outcomes: GDM (11%) and gestational hypertension (10%), chemicals were natural log-transformed and rescaled by their standard deviation (SD). Models were adjusted for serum lipids, and then adjusted for age, body mass index, race, and smoking. Models were additionally adjusted for the sum of the remaining POPs in each chemical class. Women's serum concentration of PBDE congener 153 (PBDE-153) was positively associated with an increased odds of GDM per SD increase in log-transformed concentration, for unadjusted (OR=1.36, 95%CI: 1.02-1.81), a priori adjusted (OR=1.38, 95% CI: 1.03-1.86) and with the sum of remaining PBDEs (OR=1.79, 95% CI: 1.18, 2.74) models. Our findings suggest that at environmentally relevant concentrations, maternal exposure to POPs prior to conception may contribute to increased chance of developing GDM.

Pre-pregnancy maternal exposure to polybrominated and polychlorinated biphenyls and gestational diabetes: a prospective cohort study.

BACKGROUND: While several studies have shown an association between environmental pollutants and diabetes among non-pregnant adults, few studies have prospectively assessed the association among pregnant women. We estimated the association between maternal pre-pregnancy levels of a polybrominated biphenyl (PBB 153) and 36 polychlorinated biphenyls (PCBs) with gestational diabetes (GDM). METHODS: Data are from women (18-40 years) participating in a prospective cohort who achieved pregnancy lasting ≥24 weeks gestation and completed monthly pregnancy journals (n = 258). Women were recruited between 2005 and 2007 from 16 counties in Michigan and Texas. Women who ever reported a physician diagnosis of high blood glucose in monthly pregnancy journals were categorized as having gestational diabetes (n = 28; 10.9 %). Multivariable binary logistic regression was used to estimate odds ratios (OR) and 95 % confidence intervals (CIs). RESULTS: There was no association between PBB 153 and GDM or any of the PCB congeners and GDM in unadjusted models. All associations remained non-significant with stepwise adjustment for age and waist-to-height ratio. Only with further adjustment for total serum lipids did the associations become significant, with lower levels of nine PCB congeners associated with GDM: 138, 153, 156, 167, 170, 172, 178, 180, and 194. The adjusted ORs for PCBs 170 and 180 were the strongest: 0.40 (0.18, 0.88) and 0.41 (0.19, 0.87), respectively. CONCLUSIONS: Pre-pregnancy levels of PCBs were not consistently associated with development of GDM in this prospective cohort of U.S. women. Interestingly, we found that although women with GDM had higher mean pre-pregnancy circulating lipid levels compared to women without GDM, they had lower wet weight circulating levels of many PCBs. More research is needed to understand the dynamic fluctuations of PCBs and other persistent organic pollutants between lipid compartments in women preparing to conceive and throughout pregnancy.

Parental urinary biomarkers of preconception exposure to bisphenol A and phthalates in relation to birth outcomes.

BACKGROUND: Bisphenol A (BPA) and phthalates are ubiquitous non-persistent endocrine disrupting chemicals whose relation with infant birth size is not clearly understood. METHODS: We examined associations between maternal and paternal preconception urinary concentrations of total BPA and 14 phthalate metabolites and birth size for 233 infants. Multiple linear regression models were used to estimate parental quartiles of BPA and phthalates in relation to birth weight, length, head circumference, and ponderal index with separate models run for each parent adjusting for age, smoking, body mass index, education, alcohol, parity, and creatinine. Models also included an interaction term for each chemical and infant sex and were further adjusted to include the other partner's chemical concentrations. RESULTS: In maternal models adjusted for partner's exposure and covariates, reductions in birth weight (range: 178-215 g; p < 0.05) were observed for the 2nd quartile of maternal monomethyl phthalate, mono-[(2-carboxymethyl) hexyl] phthalate and mono-n-octyl phthalate when compared with the 1st quartiles. The 3rd quartile of monoethylhexyl phthalate (mEHP) was also associated with a 200.16 g (95 % CI: -386.90, -13.42) reduction. Similar reductions in birth weight were observed for the 2(nd) quartile of paternal mEHP (ß = -191.93 g; 95 % CI: -381.61, -2.25). Additionally, select maternal urinary metabolites were associated with decreased head circumference, birth length and gestational age. However, paternal concentrations were generally associated with increased birth length and gestational age. CONCLUSIONS: We observed some suggestion that preconception maternal and paternal urinary concentration of BPA and specific phthalate metabolites may be associated with smaller birth size and increased gestational age, though the findings appeared to be parent and chemical specific.

Clustering of retrospectively reported and prospectively observed time-to-pregnancy.

PURPOSE: Given reportedly high clustering but limited validity of retrospectively reported time-to-pregnancy (TTP), we assessed within-woman clustering for retrospectively reported TTPs alone and including gold-standard prospectively observed TTPs among women with 2 or more retrospectively reported and 1 or more prospectively observed TTPs. We further investigated whether past trying times inform future trying time among women with 1 or more retrospectively reported and 1 or more prospectively observed TTPs. METHODS: Five hundred one couples attempting pregnancy were prospectively observed until human chorionic gonadotropin pregnancy or 12 months of trying. Women reported TTP for past planned pregnancies. Clustering as measured by the frailty variance was estimated using discrete Cox frailty models, adjusted for age, body mass index, smoking at each attempt. Utility of past attempts to inform future attempts was assessed with discrete Cox models and relative risk regression, adjusted for enrollment age, body mass index, smoking. RESULTS: Seventy-five women with 2 or more prior pregnancies contributed 180 retrospective and 91 prospective TTPs for frailty modeling. Retrospectively reported TTP clustering was high (frailty variance = 0.89) but substantially lower when including prospectively observed TTPs (frailty variance = 0.42). Among 202 women with 1 or more prior pregnancies, past trying times did not inform future trying time. CONCLUSIONS: TTP recall rather than TTP may account for clustering. Past trying times may not inform future trying times.

Birth outcomes and background exposures to select elements, the Longitudinal Investigation of Fertility and the Environment (LIFE).

Evidence suggests that trace exposures to select elements may increase the risk for adverse birth outcomes. To investigate further, we used multiple regression to assess associations between preconception parental exposures to Pb, Cd, and total Hg in blood, and 21 elements in urine, with n=235 singleton birth outcomes, adjusted for confounders and partner's exposure. Earlier gestational age at delivery (GA) was associated with higher tertiles of urine maternal W (-1.22 days) and paternal U (-1.07 days), but GA was later for higher tertiles of maternal (+1.11 days) and paternal (+1.30 days) blood Hg. Additional analysis indicated shorter GA associated with higher paternal urine Ba, W, and U, and with higher maternal blood Pb for boys, but GA was longer in association with higher maternal urine Cr. Birth weight (BW) was lower for higher tertiles of paternal urine Cs (-237.85g), U (-187.34g), and Zn (-209.08g), and for higher continuous Cr (P=0.021). In contrast, BW was higher for higher tertiles of paternal urine As (+194.71g) and counterintuitively for maternal blood Cd (+178.52g). Birth length (BL) was shorter for higher tertiles of urine maternal W (-1.22cm) and paternal U (-1.10cm). Yet, higher tertiles of maternal (+1.11cm) and paternal (+1.30) blood Hg were associated with longer BL. Head circumference at delivery was lower for higher tertiles of paternal urine U (-0.83cm), and for higher continuous Mo in boys (-0.57cm). Overall, associations were most consistently indicated for GA and measures of birth size with urine W and U, and paternal exposures were more frequently associated than maternal. Though limited by several factors, ours is the largest multi-element investigation of prospective couple-level trace exposures and birth outcomes to date; the novel observations for W and U merit further investigation.

Higher urinary lignan concentrations in women but not men are positively associated with shorter time to pregnancy.

Phytoestrogens have been associated with subtle hormonal changes, although effects on fecundity are unknown. Our objective was to evaluate the association between male and female urinary phytoestrogen (isoflavone and lignan) concentrations and time to pregnancy (TTP) in a population-based cohort of 501 couples desiring pregnancy and discontinuing contraception. Couples were followed for 12 mo or until pregnancy. Fecundability ORs (FORs) and 95% CIs were estimated after adjusting for age, body mass index, race, site, creatinine, supplement use, and physical activity in relation to female, male, and joint couple concentrations. Models included the phytoestrogen of interest and the sum of the remaining individual phytoestrogens. FORs <1 denote a longer TTP and FORs >1 a shorter TTP. Urinary lignan concentrations were higher, on average, among female partners of couples who became pregnant during the study compared with women who did not become pregnant (median enterodiol: 118 vs. 80 nmol/L; P < 0.10; median enterolactone: 990 vs. 412 nmol/L; P < 0.05) and were associated with significantly shorter TTP in models based on both individual and couples' concentrations (couples' models: enterodiol FOR, 1.13; 95% CI: 1.02, 1.26; enterolactone FOR, 1.11; 95% CI: 1.01, 1.21). Male lignan concentrations were not associated with TTP, nor were isoflavone concentrations. Sensitivity analyses showed that associations observed are unlikely to be explained by potential unmeasured confounding by lifestyle or other nutrients. Our results suggest that female urinary lignan concentrations at levels characteristic of the U.S. population are associated with a shorter TTP among couples who are attempting to conceive, highlighting the importance of dietary influences on fecundity.

Semen quality and time to pregnancy: the Longitudinal Investigation of Fertility and the Environment Study.

OBJECTIVE: To assess semen parameters and couple fecundity as measured by time to pregnancy (TTP). DESIGN: Observational prospective cohort with longitudinal measurement of TTP. SETTING: Sixteen Michigan/Texas counties. PATIENT(S): A total of 501 couples discontinuing contraception were followed for 1 year while trying to conceive; 473 men (94%) provided one semen sample, and 80% provided two samples. INTERVENTION(S): None. MAIN OUTCOME MEASURE(S): Using prospectively measured TTP, fecundability odds ratios (FORs) and 95% confidence intervals (CIs) were estimated for 36 individual semen quality parameters accounting for repeated semen samples, time off contraception, abstinence, enrollment site, and couples' ages, body mass indices, and serum cotinine concentrations. RESULT(S): In adjusted models, semen quality parameters were associated with significantly shorter TTP as measured by FORs >1: percent motility, strict and traditional morphology, sperm head width, elongation factor, and acrosome area. Significantly longer TTPs or FORs <1 were observed for morphologic categories amorphous and round sperm heads and neck/midpiece abnormalities. No semen quality parameters achieved significance when simultaneously modeling all other significant semen parameters and covariates, except for percent coiled tail when adjusting for sperm concentration (FOR 0.99; 95% CI 0.99-1.00). Male age was consistently associated with reduced couple fecundity (FOR 0.96; 95% CI 0.93-0.99), reflecting a longer TTP across all combined models. Female but not male body mass index also conferred a longer TTP (FOR 0.98; 95% CI 0.96-0.99). CONCLUSION(S): Several semen measures were significantly associated with TTP when modeled individually but not jointly and in the context of relevant couple-based covariates.

Strength of default mode resting-state connectivity relates to white matter integrity in children.

A default mode network of brain regions is known to demonstrate coordinated activity during the resting state. While the default mode network is well characterized in adults, few investigations have focused upon its development. We scanned 9-13-year-old children with diffusion tensor imaging and resting-state functional magnetic resonance imaging. We identified resting-state networks using Independent Component Analysis and tested whether the functional connectivity between the medial prefrontal cortex (mPFC) and posterior cingulate cortex (PCC) depends upon the maturation of the underlying cingulum white matter tract. To determine the generalizability of this relationship, we also tested whether functional connectivity depends on white matter maturity between bilateral lateral prefrontal cortex (lateral PFC) within the executive control network. We found a positive relationship between mPFC-PCC connectivity and fractional anisotropy of the cingulum bundle; this positive relationship was moderated by the age of the subjects such that it was stronger in older children. By contrast, no such structure-function relationship emerged between right and left lateral PFC. However, functional and structural connectivity of this tract related positively with cognitive speed, fluency, and set-switching neuropsychological measures.

Document classification for mining host pathogen protein-protein interactions.

OBJECTIVE: Scientific findings regarding human pathogens and their host responses are buried in the growing volume of biomedical literature and there is an urgent need to mine information pertaining to pathogenesis-related proteins especially host pathogen protein-protein interactions (HP-PPIs) from literature. METHODS: In this paper, we report our exploration of developing an automated system to identify MEDLINE abstracts referring to HP-PPIs. An annotated corpus consisting of 1360 MEDLINE abstracts was generated. With this corpus, we developed and evaluated document classification systems using support vector machines (SVMs). We also investigated the effects of three feature selection methods:information gain (IG), chi(2) test, and specific mutual information (SI). The performance was measured using normalized discounted cumulative gain (NDCG) and positive predictive value (PPV) and all measures were obtained through 10-fold cross validation. RESULTS: NDCG measures for classification systems using all features or a subset of features selected using IG and chi(2) test range from 0.83 to 0.89 while classification systems built based on features selected using SI had relatively lower NDCG measures. The classification system achieved a PPV of 50.7% for the top 10% ranked documents comparing to a baseline PPV of 10.0%. CONCLUSIONS: Our results indicate that document classification systems can be constructed to efficiently retrieve HP-PPI related documents. Feature selection was effective in reducing the dimensionality of features to build a compact system.

A meta-analysis of functional neuroimaging studies of dyslexia.

Reading and phonological processing deficits have been the primary focus of neuroimaging studies addressing the neurologic basis of developmental dyslexia, but to date there has been no objective assessment of the consistency of these findings. To address this issue, spatial coordinates reported in the literature were submitted to two parallel activation likelihood estimate (ALE) meta-analyses. First, a meta-analysis including 96 foci from nine publications identified regions where typical readers are likely to show greater activation than dyslexics: two left extrastriate areas within BA 37, precuneus, inferior parietal cortex, superior temporal gyrus, thalamus, and left inferior frontal gyrus. Right hemisphere ALE foci representing hypoactivity in dyslexia were found in the fusiform, postcentral, and superior temporal gyri. To identify regions in which dyslexic subjects reliably show greater activation than controls, 75 foci from six papers were entered into a second meta-analysis. Here ALE results revealed hyperactivity associated with dyslexia in right thalamus and anterior insula. These findings suggest that during the performance of a variety of reading tasks, normal readers activate left-sided brain areas more than dyslexic readers do, whereas dyslexia is associated with greater right-sided brain activity. The most robust result was in left extrastriate cortex, where hypoactivity associated with dyslexia was found. However, the ALE maps provided no support for cerebellar dysfunction, nor for hyperactivity in left frontal cortex in dyslexia, suggesting that these findings, unlike those described above, are likely to be more varied in terms of their reproducibility or spatial location.