Kellie Archer, PhD
College of Public Health
Molecular Biology and Cancer Genetics
I am a member of the Molecular Biology and Cancer Genetics Program. My primary research area has been in the development of statistical methods and software for analyzing data arising from high-throughput genomic experiments, which yield data sets consisting of a large number of candidate predictors (P) for a relatively small number of observations (N). Traditional statistical methods cannot be applied directly when P>N therefore alternative methodologies and software need to be developed. My methodological work has focused on developing machine learning and penalized models for ordinal and discrete response prediction. Examples of ordinal outcomes include response evaluated by RECIST criteria (complete response, partial response, stable disease, progressive disease), TNM stage, Knodell hepatic activity index. Examples of discrete (or count) outcomes, which take on a skewed rather than a Gaussian distribution, include micronuclei frequency, number of positive lymph nodes, length of stay. Another specific area of my research has pertained to quality assessment techniques for genomic assays.