Research Assistant Professor
Dr. Thomas Richards earned a doctorate in Biostatistics from the University of Pittsburgh in 2002. As part of his doctoral training he participated in the design and analysis of clinical and basic research studies of lung diseases. He worked for eight years as a Biostatistician at the University of Pittsburgh Cancer Institute, designing and analyzing research studies and clinical trials in melanoma, oral cancer, and lung cancer. He trained in bioinformatics and genomic data analysis as a research associate at The Dorothy P. & Richard P. Simmons Center for Interstitial Lung Disease. Dr. Richards is interested in finding reliable biomarkers of disease activity and/or prognosis in Idiopathic Pulmonary Fibrosis and other lung diseases.
As a Biostatistician Dr. Richards works 100% in research and his main interest is in participating as a team member in research activities that both inform and improve patient care.
Dr. Richard’s research centers on the application of biostatistical methods to the study of lung diseases; in particular, he focuses on data analytic techniques for high-throughput genomics and proteomics data, including DNA microarrays and bead-based proteomics assays. He is interested in developing novel statistical models of biological interactions involving thousands of genomic variables that jointly determine a biological response. In terms of clinical applications, he is working to determine peripheral blood biomarkers for diagnosis and prognosis in lung diseases such as Idiopathic Pulmonary Fibrosis (IPF) and Chronic Obstructive Pulmonary Disorder (COPD).
The Classification tree below was obtained by CART applied to plasma protein concentration data from IPF patients and controls. A blue box identifies a terminal node as Control; A red box as IPF. All counts are listed as Control / IPF. Concentrations are in ng/ml. In the subgroup with high MMP7 concentration but low MMP1 concentration (14 IPF samples, 5 control samples), splitting on IGFBP1 and TNFRSF1A improves classification, while in the subgroup with low MMP7, MMP8 improves classification.
The graph to the left shows ROC curves for using each of five markers, or their combination, to classify samples as IPF or control. Sensitivity, or true positive rate, is plotted on the y-axis and false positive rate, or 1 minus specificity, on the x-axis. The area under each ROC curve is equivalent to the numerator of the Mann-Whitney U statistic comparing the marker distributions between IPF and control samples. The magenta line labeled "Combined" is for the combinatorial classifier using all five markers. The identity line at 45 degrees represents a marker that performed no better than classifying samples as IPF or Control by flipping a fair coin.
Rosas IO*, Richards TJ*, Konishi K, Zhang Y, Gibson, K, Lokshin AE, Lindell, KO, Cisneros, J, MacDonald, S, Pardo, A, Sciurba F, Dauber, J, Selman, M, Gochuico BR, Kaminski N. MMP1 and MMP7 as Potential Peripheral Blood Biomarkers in Idiopathic Pulmonary Fibrosis. 2008 PLoS Med Apr 29;5(4):e93. (*These authors contributed equally.)
Selman M, Carrillo G, Estrada A, Mejia M, Becerril C, Cisneros J, Gaxiola M, Perez-Padilla R, Navarro C, Richards T, Dauber J, King TE, Pardo A, Kaminski N. Accelerated Variant of Idiopathic Pulmonary Fibrosis: Clinical Behavior and Gene Expression Pattern. PLOS One 2007 2(5): e482.
Pardo A, Gibson K, Cisneros J, Richards T, Yang Y, Yousem S, Becerril C, Herrera I, Ruiz V, Selman M, Kaminski N. Upregulation and profibrotic role of osteopontin in human idiopathic pulmonary fibrosis. PLoS Medicine 2005 Sep;2(9):e251.