Partial least square (PLS) methods (also sometimes called projection to latent structures) relate the information present in two data tables that collect measurements on the same set of observations. ...PLS methods proceed by deriving latent variables which are (optimal) linear combinations of the variables of a data table. When the goal is to find the shared information between two tables, the approach is equivalent to a correlation problem and the technique is then called partial least square correlation (PLSC) (also sometimes called PLS-SVD). In this case there are two sets of latent variables (one set per table), and these latent variables are required to have maximal covariance. When the goal is to predict one data table the other one, the technique is then called partial least square regression. In this case there is one set of latent variables (derived from the predictor table) and these latent variables are required to give the best possible prediction. In this paper we present and illustrate PLSC and PLSR and show how these descriptive multivariate analysis techniques can be extended to deal with inferential questions by using cross-validation techniques such as the bootstrap and permutation tests.
Partial Least Squares (PLS) methods are particularly suited to the analysis of relationships between measures of brain activity and of behavior or experimental design. In neuroimaging, PLS refers to ...two related methods: (1) symmetric PLS or Partial Least Squares Correlation (PLSC), and (2) asymmetric PLS or Partial Least Squares Regression (PLSR). The most popular (by far) version of PLS for neuroimaging is PLSC. It exists in several varieties based on the type of data that are related to brain activity: behavior PLSC analyzes the relationship between brain activity and behavioral data, task PLSC analyzes how brain activity relates to pre-defined categories or experimental design, seed PLSC analyzes the pattern of connectivity between brain regions, and multi-block or multi-table PLSC integrates one or more of these varieties in a common analysis. PLSR, in contrast to PLSC, is a predictive technique which, typically, predicts behavior (or design) from brain activity. For both PLS methods, statistical inferences are implemented using cross-validation techniques to identify significant patterns of voxel activation. This paper presents both PLS methods and illustrates them with small numerical examples and typical applications in neuroimaging.
Multiple factor analysis (MFA, also called multiple factorial analysis) is an extension of principal component analysis (PCA) tailored to handle multiple data tables that measure sets of variables ...collected on the same observations, or, alternatively, (in dual‐MFA) multiple data tables where the same variables are measured on different sets of observations. MFA proceeds in two steps: First it computes a PCA of each data table and ‘normalizes’ each data table by dividing all its elements by the first singular value obtained from its PCA. Second, all the normalized data tables are aggregated into a grand data table that is analyzed via a (non‐normalized) PCA that gives a set of factor scores for the observations and loadings for the variables. In addition, MFA provides for each data table a set of partial factor scores for the observations that reflects the specific ‘view‐point’ of this data table. Interestingly, the common factor scores could be obtained by replacing the original normalized data tables by the normalized factor scores obtained from the PCA of each of these tables. In this article, we present MFA, review recent extensions, and illustrate it with a detailed example. WIREs Comput Stat 2013, 5:149–179. doi: 10.1002/wics.1246
This article is categorized under:
Data: Types and Structure > Categorical Data
Statistical Learning and Exploratory Methods of the Data Sciences > Exploratory Data Analysis
Statistical and Graphical Methods of Data Analysis > Multivariate Analysis
Depression and chronic pain are the two most important causes of disability (Global Burden of Disease Study 2013). They occur together more frequently than expected and both conditions have been ...shown to be co-morbid with cardiovascular disease. Although shared socio-demographic risk factors (e.g. gender, deprivation) might explain the co-morbidity of these three conditions, we hypothesised that these three long-term, highly prevalent conditions co-occur and may be due to shared familial risk, and/or genetic factors.
We employed three different study designs in two independent cohorts, namely Generation Scotland and TwinsUK, having standardised, validated questionnaire data on the three traits of interest. First, we estimated the prevalence and co-occurrence of chronic pain, depression and angina among 24,024 participants of a population-based cohort of extended families (Generation Scotland: Scottish Family Health Study), adjusting for age, gender, education, smoking status, and deprivation. Secondly, we compared the odds of co-morbidity in sibling-pairs with the odds in unrelated individuals for the three conditions in the same cohort. Lastly, examination of similar traits in a sample of female twins (TwinsUK, n = 2,902), adjusting for age and BMI, allowed independent replication of the findings and exploration of the influence of additive genetic (A) factors and shared (C) and non-shared (E) environmental factors predisposing to co-occurring chronic widespread pain (CWP) and cardiovascular disease (hypertension, angina, stroke, heart attack, elevated cholesterol, angioplasty or bypass surgery). In the Generation Scotland cohort, individuals with depression were more than twice as likely to have chronic pain as those without depression (adjusted OR 2·64 95% CI 2·34-2·97); those with angina were four times more likely to have chronic pain (OR 4·19 3·64-4·82); those with depression were twice as likely to have angina (OR 2·20 1·90-2·54). Similar odds were obtained when the outcomes and predictors were reversed and similar effects seen among sibling pairs; depression in one sibling predicted chronic pain in the other (OR 1·34 1·05-1·71), angina predicted chronic pain in the other (OR 2·19 1·63-2·95), and depression, angina (OR 1·98 1·49-2·65). Individuals with chronic pain and angina showed almost four-fold greater odds of depression compared with those manifesting neither trait (OR 3·78 2·99-4·78); angina showed seven-fold increased odds in the presence of chronic pain and depression (OR 7·76 6·05-9·95) and chronic pain nine-fold in the presence of depression and angina (OR 9·43 6·85-12·98). In TwinsUK, the relationship between CWP and depression has been published (R = 0.34, p<0.01). Considering the CWP-cardiovascular relationship, the most suitable model to describe the observed data was a combination of A, C and E, with a small but significant genetic predisposition, shared between the two traits (2·2% 95% CI 0·06-0·23).
We found an increased co-occurrence of chronic pain, depression and cardiovascular disease in two independent cohorts (general population-based cohort, twins cohort) suggesting a shared genetic contribution. Adjustment for known environmental influences, particularly those relating to socio-economic status (Generation Scotland: age, gender, deprivation, smoking, education; Twins UK: age,BMI) did not explain the relationship observed between chronic pain, depression and cardiovascular disease. Our findings from two independent cohorts challenge the concept of traditional disease boundaries and warrant further investigation of shared biological mechanisms.
Compared to the abundance of laboratory-based memory tasks, few measures exist to assess self-reported memory function. This need is particularly important for naturalistic mnemonic capacities, such ...as autobiographical memory (recall of events and facts from one's past), because it is difficult to reliably assess in the laboratory. Furthermore, naturalistic mnemonic capacities may show stable individual differences that evade the constraints of laboratory testing. The Survey of Autobiographical Memory (SAM) was designed to assess such trait mnemonics, or the dimensional characterization of self-reported mnemonic characteristics. The SAM comprises items assessing self-reported episodic autobiographical, semantic, and spatial memory, as well as future prospection. In a large sample of healthy young adults, the latent dimensional structure of the SAM was characterized with multiple correspondence analysis (MCA). This analysis revealed dimensions corresponding to general mnemonic abilities (i.e., good vs poor memory across subtypes), spatial memory, and future prospection. While episodic and semantic items did not separate in this data-driven analysis, these categories did show expected dissociations in relation to depression history and to laboratory-based measures of recollection. Remote spatial memory as assessed by the SAM showed the expected advantage for males over females. Spatial memory was also related to autobiographical memory performance. Brief versions of the SAM are provided for efficient research applications. Individual differences in memory function are likely related to other health-related factors, including personality, psychopathology, dementia risk, brain structure and function, and genotype. In conjunction with laboratory or performance based assessments, the SAM can provide a useful measure of naturalistic self-report trait mnemonics for probing these relationships.
The correct estimation of fibre orientations is a crucial step for reconstructing human brain tracts. Bayesian Estimation of Diffusion Parameters Obtained using Sampling Techniques (bedpostx) is able ...to estimate several fibre orientations and their diffusion parameters per voxel using Markov Chain Monte Carlo (MCMC) in a whole brain diffusion MRI data, and it is capable of running on GPUs, achieving speed-up of over 100 times compared to CPUs. However, few studies have looked at whether the results from the CPU and GPU algorithms differ. In this study, we compared CPU and GPU bedpostx outputs by running multiple trials of both algorithms on the same whole brain diffusion data and compared each distribution of output using Kolmogorov-Smirnov tests.
We show that distributions of fibre fraction parameters and principal diffusion direction angles from bedpostx and bedpostx_gpu display few statistically significant differences in shape and are localized sparsely throughout the whole brain. Average output differences are small in magnitude compared to underlying uncertainty.
Despite small amount of differences in output between CPU and GPU bedpostx algorithms, results are comparable given the difference in operation order and library usage between CPU and GPU bedpostx.
Sensitive periods in human development have often been proposed to explain age-related differences in the attainment of a number of skills, such as a second language (L2) and musical expertise. It is ...difficult to reconcile the negative consequence this traditional view entails for learning after a sensitive period with our current understanding of the brain's ability for experience-dependent plasticity across the lifespan. What is needed is a better understanding of the mechanisms underlying auditory learning and plasticity at different points in development. Drawing on research in language development and music training, this review examines not only what we learn and when we learn it, but also how learning occurs at different ages. First, we discuss differences in the mechanism of learning and plasticity during and after a sensitive period by examining how language exposure versus training forms language-specific phonetic representations in infants and adult L2 learners, respectively. Second, we examine the impact of musical training that begins at different ages on behavioral and neural indices of auditory and motor processing as well as sensorimotor integration. Third, we examine the extent to which childhood training in one auditory domain can enhance processing in another domain via the transfer of learning between shared neuro-cognitive systems. Specifically, we review evidence for a potential bi-directional transfer of skills between music and language by examining how speaking a tonal language may enhance music processing and, conversely, how early music training can enhance language processing. We conclude with a discussion of the role of attention in auditory learning for learning during and after sensitive periods and outline avenues of future research.
Fetal programming is a conceptual framework whereby the in utero environment shapes the offspring's neurodevelopment. Maternal depression and treatment with selective serotonin reuptake inhibitor ...(SSRI) antidepressants during pregnancy are common prenatal exposures that affect critical early life developmental programming processes. Prenatal depression and SSRIs both have been reported to increase the risks for preterm birth, low birth weight, and are associated with behavioral disturbances across the early life span. However, not all exposures lead to adverse developmental outcomes and distinguishing how each exposure contributes to variations in development remains challenging. Advances in neuroimaging, using MR and EEG, offer novel insights into central processes that might reveal the neural correlates of fetal programming. This review focuses on emerging findings from neuroimaging studies reflecting early brain functional and structural development associated with prenatal exposure to maternal depression and SSRI antidepressants. Suggestions for future research directions that use neuroimaging as a tool to advancing our understanding of the early origins of developmental plasticity are offered.
It is well established that temporal lobe epilepsy—the most common and well-studied form of epilepsy—can impair communication by disrupting social-emotional and language functions. In pediatric ...epilepsy, where seizures co-occur with the development of critical brain networks, age of onset matters: The earlier in life seizures begin, the worse the disruption in network establishment, resulting in academic hardship and social isolation. Yet, little is known about the processes by which epileptic activity disrupts developing human brain networks. Here we take a synthetic perspective—reviewing a range of research spanning studies on molecular and oscillatory processes to those on the development of large-scale functional networks—in support of a novel model of how such networks can be disrupted by epilepsy. We seek to bridge the gap between research on molecular processes, on the development of human brain circuitry, and on clinical outcomes to propose a model of how epileptic activity disrupts brain development.
Prenatal exposure to selective serotonin reuptake inhibitor (SSRI) antidepressants may influence white matter (WM) development, as previous studies report widespread microstructural alterations and ...reduced interhemispheric connectivity in SSRI‐exposed infants. In rodents, perinatal SSRIs had sex‐specific disruptions in corpus callosum (CC) axon architecture and connectivity; yet it is unknown whether SSRI‐related brain outcomes in humans are sex specific. In this study, the neonate CC was selected as a region‐of‐interest to investigate whether prenatal SSRI exposure has sex‐specific effects on early WM microstructure. On postnatal day 7, diffusion tensor imaging was used to assess WM microstructure in SSRI‐exposed (n = 24; 12 male) and nonexposed (n = 48; 28 male) term‐born neonates. Fractional anisotropy was extracted from CC voxels and a multivariate discriminant analysis was used to identify latent patterns differing between neonates grouped by SSRI‐exposure and sex. Analysis revealed localized variations in CC fractional anisotropy that significantly discriminated neonate groups and correctly predicted group membership with an 82% accuracy. Such effects were identified across three dimensions, representing sex differences in SSRI‐exposed neonates (genu, splenium), SSRI‐related effects independent of sex (genu‐to‐rostral body), and sex differences in nonexposed neonates (isthmus‐splenium, posterior midbody). Our findings suggest that CC microstructure may have a sex‐specific, localized, developmental sensitivity to prenatal SSRI exposure.