Houston Methodist. Leading Medicine.
Houston Methodist. Leading Medicine

Center for Biostatistics

To learn more or schedule resources for this Core, please login to iLab.

 

 

About

The Center for Biostatistics is a fee-for-service Core which provides consultation on study design and analyses to grant/contract-supported investigative research teams and clinical research groups. Center personnel participate in grants, contracts, teaching, and private consultations. Altogether, the cumulative man-years of experience among Core personnel in statistics, mathematics, computation, and data management exceeds 50 years.  Core personnel specialize in research project design and conceptualization, development of goals and specific aims and hypotheses, missing data imputation, Monte Carlo uncertainty analysis for risk assessment, simulation of correlation and probability distributions, inferential hypothesis testing with parametric/non-parametric tests, univariate and multivariate modeling, longitudinal modeling.

Computational Research and Numerical Methods

Applied computational research includes machine learning and computational intelligence (artificial neural networks) for knowledge discovery, linear and nonlinear dimensional reduction using manifold learning and eigendecomposition, classification analysis, text mining.  Research pursuits focus on random matrix theory, covariance matrix filtering, power laws, fractal geometry of time series, and the Marcenko-Pastur and Tracy-Widom distributions for risk modeling of fat-tail distributions.

Hardware and statistical software

Computational resources in CFB include a networked 320-core (physical) cluster and numerous PCs. HPC data mining and pattern recognition/signal processing software includes Peltarion Synapse, Predictive Dynamix, StatSoft Data Miner, Golden Helix, and Matlab. For statistical software, CFB staff use Stata (Version 12, Windows) for the bulk of analytic analyses, and PASS 12 for power and sample size determination. SAS Base/STAT is also maintained for some procedures not offered in Stata. CFB staff has experience in SAS and SPSS and can nevertheless consult on data setup, analysis, and interpretation from these packages. Other software used by CFB staff includes Minitab, Sigmastat, Sigmaplot.

Statistical Analysis Services

  • Study design and conceptualization
  • Data collection, storage/retrieval, warehousing
  • Sample size determination and statistical power
  • Percentage points for discrete and continuous probability distributions (Binomial, Poisson, Standard Normal, Student's t, F-ratio, Laplace, Skew-Normal, Logistic, etc.)
  • Monte Carlo simulation and uncertainty analysis
  • Inferential hypothesis testing
  • Linear models (ANOVA, ANACOVA, MANOVA, regression)
  • Univariate and multivariate regression modeling and diagnostics
  • Survival and actuarial analysis
  • Longitudinal data analysis (GLM, GEE, RMANOVA)
  • Bioinformatics (microarray QC, normalization, gene profile identification, cluster, heat maps)
  • Data mining (knowledge discovery and pattern recognition in data/images, unsupervised class discovery, supervised class prediction, nonlinear dimensional reduction, duo-mining)

 

Policies and Disclaimers

Our recommendation for statistical power is always 100% (two-tailed); lower values used in design (grant applications) are at the discretion of the applicant.  We follow the Committee on Publication Ethics (COPE,http://www.publicationethics.org/) regarding self-pliagerism, recycling, and ICMJE (http://www.icmje.org/ethical_1author.html) criteria for authorship: Substantial contribution to data analyses or manuscript preparation constitutes authorship and not acknowledgment.  Authorship  cannot be exchanged with currency to pay for services (no bartering).