Caching and Visualizing Statistical Analyses

We present the cacher and CodeDepends packages for R, which provide tools for (1) caching and analyzing the code for statistical analyses and (2) distributing these analyses to others in an efficient manner over the web. The cacher package takes objects created by evaluating R expressions and stores them in key-value databases. These databases of cached objects can subsequently be assembled into “cache packages” for distribution over the web. The cacher package also provides tools to help readers examine the data and code in a statistical analysis and reproduce, modify, or improve upon the results. In addition, readers can easily conduct alternate analyses of the data. The CodeDepends package provides complementary tools for analyzing and visualizing the code for a statistical analysis and this functionality has been integrated into the cacher package. In this chapter we describe the cacher and CodeDepends packages and provide examples of how they can be used for reproducible research.

USING THE R PACKAGE crlmm FOR GENOTYPING AND COPY NUMBER ESTIMATION

Genotyping platforms such as Affymetrix can be used to assess genotype-phenotype as well as copy number-phenotype associations at millions of markers. While genotyping algorithms are largely concordant when assessed on HapMap samples, tools to assess copy number changes are more variable and often discordant. One explanation for the discordance is that copy number estimates are susceptible to systematic differences between groups of samples that were processed at different times or by different labs. Analysis algorithms that do not adjust for batch effects are prone to spurious measures of association. The R package crlmm implements a multilevel model that adjusts for batch effects and provides allele-specific estimates of copy number. This paper illustrates a workflow for the estimation of allele-specific copy number, develops markerand study-level summaries of batch effects, and demonstrates how the marker-level estimates can be integrated with complimentary Bioconductor software for inferring regions of copy number gain or loss. All analyses are performed in the statistical environment R. A compendium for reproducing the analysis is available from the author’s website (http://www.biostat.jhsph.edu/~rscharpf/crlmmCompendium/index.html).

Statistical evaluation of time-dependent metabolite concentrations: estimation of post-mortem intervals based on in situ 1H-MRS of the brain

Knowledge of the time interval from death (post-mortem interval, PMI) has an enormous legal, criminological and psychological impact. Aiming to find an objective method for the determination of PMIs in forensic medicine, 1H-MR spectroscopy (1H-MRS) was used in a sheep head model to follow changes in brain metabolite concentrations after death. Following the characterization of newly observed metabolites (Ith et al., Magn. Reson. Med. 2002; 5: 915-920), the full set of acquired spectra was analyzed statistically to provide a quantitative estimation of PMIs with their respective confidence limits. In a first step, analytical mathematical functions are proposed to describe the time courses of 10 metabolites in the decomposing brain up to 3 weeks post-mortem. Subsequently, the inverted functions are used to predict PMIs based on the measured metabolite concentrations. Individual PMIs calculated from five different metabolites are then pooled, being weighted by their inverse variances. The predicted PMIs from all individual examinations in the sheep model are compared with known true times. In addition, four human cases with forensically estimated PMIs are compared with predictions based on single in situ MRS measurements. Interpretation of the individual sheep examinations gave a good correlation up to 250 h post-mortem, demonstrating that the predicted PMIs are consistent with the data used to generate the model. Comparison of the estimated PMIs with the forensically determined PMIs in the four human cases shows an adequate correlation. Current PMI estimations based on forensic methods typically suffer from uncertainties in the order of days to weeks without mathematically defined confidence information. In turn, a single 1H-MRS measurement of brain tissue in situ results in PMIs with defined and favorable confidence intervals in the range of hours, thus offering a quantitative and objective method for the determination of PMIs.

Utilizing Gaussian-Schell model beams to mitigate atmospheric turbulence in free space optical communications / by Kyle R. Drexler.

Turbulence affects traditional free space optical communication by causing speckle to appear in the received beam profile. This occurs due to changes in the refractive index of the atmosphere that are caused by fluctuations in temperature and pressure, resulting in an inhomogeneous medium. The Gaussian-Schell model of partial coherence has been suggested as a means of mitigating these atmospheric inhomogeneities on the transmission side. This dissertation analyzed the Gaussian-Schell model of partial coherence by verifying the Gaussian-Schell model in the far-field, investigated the number of independent phase control screens necessary to approach the ideal Gaussian-Schell model, and showed experimentally that the Gaussian-Schell model of partial coherence is achievable in the far-field using a liquid crystal spatial light modulator. A method for optimizing the statistical properties of the Gaussian-Schell model was developed to maximize the coherence of the field while ensuring that it does not exhibit the same statistics as a fully coherent source. Finally a technique to estimate the minimum spatial resolution necessary in a spatial light modulator was developed to effectively propagate the Gaussian-Schell model through a range of atmospheric turbulence strengths. This work showed that regardless of turbulence strength or receiver aperture, transmitting the Gaussian-Schell model of partial coherence instead of a fully coherent source will yield a reduction in the intensity fluctuations of the received field. By measuring the variance of the intensity fluctuations and the received mean, it is shown through the scintillation index that using the Gaussian-Schell model of partial coherence is a simple and straight forward method to mitigate atmospheric turbulence instead of traditional adaptive optics in free space optical communications.

Statistical analysis of multichannel scalp field data

High density spatial and temporal sampling of EEG data enhances the quality of results of electrophysiological experiments. Because EEG sources typically produce widespread electric fields (see Chapter 3) and operate at frequencies well below the sampling rate, increasing the number of electrodes and time samples will not necessarily increase the number of observed processes, but mainly increase the accuracy of the representation of these processes. This is namely the case when inverse solutions are computed. As a consequence, increasing the sampling in space and time increases the redundancy of the data (in space, because electrodes are correlated due to volume conduction, and time, because neighboring time points are correlated), while the degrees of freedom of the data change only little. This has to be taken into account when statistical inferences are to be made from the data. However, in many ERP studies, the intrinsic correlation structure of the data has been disregarded. Often, some electrodes or groups of electrodes are a priori selected as the analysis entity and considered as repeated (within subject) measures that are analyzed using standard univariate statistics. The increased spatial resolution obtained with more electrodes is thus poorly represented by the resulting statistics. In addition, the assumptions made (e.g. in terms of what constitutes a repeated measure) are not supported by what we know about the properties of EEG data. From the point of view of physics (see Chapter 3), the natural “atomic” analysis entity of EEG and ERP data is the scalp electric field

Coping resources in a sample of chronic low back pain patients: Evaluation of the questionnaire for back pain

BACKGROUND The coping resources questionnaire for back pain (FBR) uses 12 items to measure the perceived helpfulness of different coping resources (CRs, social emotional support, practical help, knowledge, movement and relaxation, leisure and pleasure, spirituality and cognitive strategies). The aim of the study was to evaluate the instrument in a clinical patient sample assessed in a primary care setting. SAMPLE AND METHODS The study was a secondary evaluation of empirical data from a large cohort study in general practices. The 58 participating primary care practices recruited patients who reported chronic back pain in the consultation. Besides the FBR and a pain sketch, the patients completed scales measuring depression, anxiety, resilience, sociodemographic factors and pain characteristics. To allow computing of retested parameters the FBR was sent to some of the original participants again after 6 months (90% response rate). We calculated consistency and retest reliability coefficients as well as correlations between the FBR subscales and depression, anxiety and resilience scores to account for validity. By means of a cluster analysis groups with different resource profiles were formed. Results. RESULTS For the study 609 complete FBR baseline data sets could be used for statistical analysis. The internal consistency scores ranged fromα=0.58 to α=0.78 and retest reliability scores were between rTT=0.41 and rTT=0.63. Correlation with depression, fear and resilience ranged from r=-0.38 to r=0.42. The cluster analysis resulted in four groups with relatively homogenous intragroup profiles (high CRs, low spirituality, medium CRs, low CRs). The four groups differed significantly in fear and depression (the more inefficient the resources the higher the difference) as well as in resilience (the more inefficient the lower the difference). The group with low CRs also reported permanent pain with no relief. The groups did not otherwise differ. CONCLUSIONS The FBR is an economic instrument that is suitable for practical use e.g. in primary care practices to identify strengths and deficits in the CRs of chronic pain patients that can then be specified in face to face consultation. However, due to the rather low reliability, the use of subscales for profile differentiation and follow-up measurement in individual diagnoses is limited.

Are clustering effects accounted for in statistical analysis in leading dental specialty journals?

OBJECTIVES In dental research multiple site observations within patients or taken at various time intervals are commonplace. These clustered observations are not independent; statistical analysis should be amended accordingly. This study aimed to assess whether adjustment for clustering effects during statistical analysis was undertaken in five specialty dental journals. METHODS Thirty recent consecutive issues of Orthodontics (OJ), Periodontology (PJ), Endodontology (EJ), Maxillofacial (MJ) and Paediatric Dentristry (PDJ) journals were hand searched. Articles requiring adjustment accounting for clustering effects were identified and statistical techniques used were scrutinized. RESULTS Of 559 studies considered to have inherent clustering effects, adjustment for this was made in the statistical analysis in 223 (39.1%). Studies published in the Periodontology specialty accounted for clustering effects in the statistical analysis more often than articles published in other journals (OJ vs. PJ: OR=0.21, 95% CI: 0.12, 0.37, p<0.001; MJ vs. PJ: OR=0.02, 95% CI: 0.00, 0.07, p<0.001; PDJ vs. PJ: OR=0.14, 95% CI: 0.07, 0.28, p<0.001; EJ vs. PJ: OR=0.11, 95% CI: 0.06, 0.22, p<0.001). A positive correlation was found between increasing prevalence of clustering effects in individual specialty journals and correct statistical handling of clustering (r=0.89). CONCLUSIONS The majority of studies in 5 dental specialty journals (60.9%) examined failed to account for clustering effects in statistical analysis where indicated, raising the possibility of inappropriate decreases in p-values and the risk of inappropriate inferences.

Coronary artery stents: influence of adaptive statistical iterative reconstruction on image quality using 64-HDCT

OBJECTIVE: The assessment of coronary stents with present-generation 64-detector row computed tomography (HDCT) scanners is limited by image noise and blooming artefacts. We evaluated the performance of adaptive statistical iterative reconstruction (ASIR) for noise reduction in coronary stent imaging with HDCT. METHODS AND RESULTS: In 50 stents of 28 patients (mean age 64 ± 10 years) undergoing coronary CT angiography (CCTA) on an HDCT scanner the mean in-stent luminal diameter, stent length, image quality, in-stent contrast attenuation, and image noise were assessed. Studies were reconstructed using filtered back projection (FBP) and ASIR-FBP composites. ASIR resulted in reduced image noise vs. FBP (P < 0.0001). Two readers graded the CCTA stent image quality on a 4-point Likert scale and determined the proportion of interpretable stent segments. The best image quality for all clinical images was obtained with 40 and 60% ASIR with significantly larger luminal area visualization compared with FBP (+42.1 ± 5.4% with 100% ASIR vs. FBP alone; P < 0.0001) while the stent length was decreased (-4.7 ± 0.9%,

Analyzing diffuse scattering with supercomputers

Two new approaches to quantitatively analyze diffuse diffraction intensities from faulted layer stacking are reported. The parameters of a probability-based growth model are determined with two iterative global optimization methods: a genetic algorithm (GA) and particle swarm optimization (PSO). The results are compared with those from a third global optimization method, a differential evolution (DE) algorithm [Storn & Price (1997). J. Global Optim. 11, 341–359]. The algorithm efficiencies in the early and late stages of iteration are compared. The accuracy of the optimized parameters improves with increasing size of the simulated crystal volume. The wall clock time for computing quite large crystal volumes can be kept within reasonable limits by the parallel calculation of many crystals (clones) generated for each model parameter set on a super- or grid computer. The faulted layer stacking in single crystals of trigonal three-pointedstar- shaped tris(bicylco[2.1.1]hexeno)benzene molecules serves as an example for the numerical computations. Based on numerical values of seven model parameters (reference parameters), nearly noise-free reference intensities of 14 diffuse streaks were simulated from 1280 clones, each consisting of 96 000 layers (reference crystal). The parameters derived from the reference intensities with GA, PSO and DE were compared with the original reference parameters as a function of the simulated total crystal volume. The statistical distribution of structural motifs in the simulated crystals is in good agreement with that in the reference crystal. The results found with the growth model for layer stacking disorder are applicable to other disorder types and modeling techniques, Monte Carlo in particular.