RECENT WORK EXPERIENCE
Senior Data Scientist Dec 2016 – Present
AT&T, Chief Data Office
Collaborated with senior business leaders and technical team to identify opportunities for retail optimization; proposed methods for retail evaluation and extraction of actionable insights to drive revenue growth
Designed pilot study for testing proposed solutions in retail locations
Created visualizations to communicate study performance metrics to senior business leaders and technical team
Performed data exploration and data cleaning to facilitate development of prediction models
Developed and validated classification models using machine learning algorithms
Research Associate (Assistant Professor) Aug 2011 - Present
The University of Chicago, Biostatistics Laboratory http://bit.ly/spw-uchicago
Consulted with over 110 biomedical research teams in The University of Chicago Medical Center, designing clinical research studies (sample size calculation and analysis plan development), analyzing study data, and summarizing results for the study team and greater scientific community.
Developed a survival probability prediction equation for patients on the national lung transplant waiting list given missing covariate data and a large number of highly correlated features. Compared to other published survival prediction models for this population, survival estimates obtained using the novel approach more closely resembled Kaplan-Meier estimates at 1, 2, and 5 years.
Relevant skills/tools: bootstrap, model validation, elastic net http://bit.ly/spw-survival-prediction
Developed a missing data imputation algorithm for interval censored count covariate data used to examine the relationship between overall survival and rate of disease progression.
Relevant skills/tools: mixed effects model, joint longitudinal-survival model http://bit.ly/spw-jointmodel
Investigated the changes in levels and trends of market share of popular attention-deficit hyperactivity disorder (ADHD) treatments after three FDA public health advisories using segmented time series regression models.
Relevant skills/tools: generalized least squares regression http://bit.ly/spw-adhd
Designed and implemented a simulation study of a mixed-effects location scale model [http://bit.ly/spw-mixregls]. Significantly reduced run time by simulating data in Python (within R) and parallelizing across blocks of simulated datasets using 16 processors and 70 GB RAM on a Linux-based cluster node.
Relevant skills/tools: Python, parallel computing (manuscript in progress)
Served as Principal Biostatistician for a 5 year longitudinal biomarker study of wall shear stress as a candidate predictor for cephalic arch stenosis. Designed a customized REDCap database for mixed-mode patient study data and created customized dashboard reports.
Relevant skills/tools: survival analysis, data management http://bit.ly/spw-hammes-cas
Graduate research assistant: Los Alamos National Laboratory May 2009 - Jul 2011
Division of Theoretical Biology and Biophysics http://bit.ly/spw-hiv
Biostatistician for an international multi-disciplinary, multi-institutional team of HIV vaccine researchers testing theoretical design strategies in a remote animal research laboratory. Duties include analysis of immune response data, presentation of study findings to research team, and production of high-quality figures for team leader’s conference presentations and journal submissions.
Compared immune responses elicited from several candidate vaccine design strategies and select a subset of vaccines to be tested in the next iteration of vaccine trials.
Developed a Bayesian Poisson fixed- and mixed-effects regression models for interval censored HIV vaccine immune response counts (dissertation: http://bit.ly/2bfpJQk).