Skip to:

Richard A. Olshen

Richard A. Olshen, PhD

Professor of Biomedical Data Science, Emeritus
Stanford Health Policy Associate

Sequoia Hall, Room 228
390 Serra Mall
Stanford, CA 94305

Assistant: Bonnie Chung
bchung@stanford.edu

(650) 725-2241 (voice)
(650) 725-6951 (fax)

Research Interests

Statistical/genetic aspects of the adaptive human immune system; all manner of biostatistical inference, including but not limited to health policy

Bio

Professor Olshen is a Fellow of The Institute of Mathematical Statistics, The American Statistical Association, The American Association for the Advancement of Science, and The Institute of Electrical and Electronics Engineers.  He is an elected member of the International Statistical Institute. He has been a Guggenheim Fellow and the recipient of a Research Scholar in Cancer Award from the American Cancer Society. His interests include the development of statistical methods for prediction and the assessment of accuracy. He is one of the developers of CARTª binary tree-structured methods for classification, regression, and probability class estimation and of their extensions to survival analysis and clustering. In collaboration with others, he has studied these algorithms theoretically and has applied them to the computer-aided diagnosis of heart attack, as well as to making prognoses for patients with lymphoma, extracting features of organic compounds that tend to make them ulcerogenic, to data compression and the automated detection attempt to find the genes that predispose to hypertension, and to the definition of health states in health services research. His current research also involves the development of parsimonious models for describing longitudinal data, especially as they apply to understanding autoimmune disease of the kidney. Typically, these consist of the sum of an overall mean function and subject-specific coefficients of suitably smoothed eigenfunctions of residuals. In the past, he collaborated with Alan Garber in developing technologies for tracking cholesterol longitudinally in time and quantifying the accuracy of findings. Their ideas are now finding wide-ranging application.