The Analytics and Biostatistics Core (ABC) will provide cost efficient, responsive and integrative data management and statistical and analytic support for Center projects. It will provide easy access to data, targeted consultations and facilitate collaboration for cross disciplinary and integrative research. It will utilize, adapt and/or develop novel and efficient analytic methods and designs that allow the study of biological mechanisms leading to the facilitation of personalized medicine. It will play a strong role in integrating findings across projects using both quantitative and qualitative approaches. Dr. Eugene Laska will be the director and Dr. Carole Siegel the deputy director of the Core staffed by an additional statistical/data scientist, Dr. Meng Qian and a data management expert.
The specific aims of the ABC are to provide: 1) consultation on the details of the experimental designs of Center projects, data collection, data quality control and to maintain centralized documentation of these efforts; 2) data management and storage services enabling efficient and secured data sharing, data integration and data visualization; 3) analysis of data from each and across Center research projects applying as appropriate state-of-the-art bioinformatics, statistical modeling, machine learning, and causal analysis tools and algorithms. 4) new analytic methodologies to further inform personalized medicine. Novel computational and analytical approaches will be applied or developed including a new paradigm for the analysis of clinical trial data based on biomarkers for causal modelling of the probability of treatment response that can serve in its application to move the field towards more personalized medicine. ABC staff are well versed in state of the art statistical methods for analyzing data including ANOVAs, mixed models and the application of regression models, trajectory analysis to examine variation over time, latent variables in their use in factor analysis and survival time methods in their use to examine onset and relapse. They have experience in handling unbalanced samples in terms of potentially prognostic measures, adjusting for batch effects, controlling for confounding variables including demographics, comorbidities and health conditions. ABC staff has proficiency in the use of data analytic/mining techniques for classification and clustering most particularly random forests and in methods for identifying important variables in in regression such as lasso, ridge regression and elastic net regression.
