KEITH HUMPHREYS

Natural history models of breast cancer  - Understanding disease progression and screening sensitivity

Zoom link: https://zoom.us/j/98120936054

Abstract
Despite dramatic clinical achievements improving prognosis, 20% of breast cancer patients die from their disease. Breast tumors are very variable in their natural histories (rates of growth and spread) and in terms of prognosis. It is possible that population-screening programs could be more efficient, and there are large ongoing, and planned, studies of personalized screening based on risk and mammographic/breast density (amount of fibrous connective or glandular tissue). More needs to be understood about the factors behind an individual's risk of aggressive cancer, the individual-specific probabilities of tumors being detected at screening and the chances of responding to treatment. We have been developing biologically motivated, random effects models of tumour progression and using them to analyse data collected in large epidemiological studies of breast cancer, with information on screening.  In this talk I will describe some of the issues we have encountered in developing these approaches for cohort studies and for retrospective studies of incident cases. 

Bio
Keith Humphreys was awarded a doctoral degree in Statistics at Southampton University (U.K.) in 1996. In 2005, he was appointed Senior Lecturer in Biostatistics at Karolinska Institutet, Stockholm, Sweden, at the Department of Medical Epidemiology and Biostatistics. In 2019, he was appointed as Professor in Biostatistics in the same department. His research interests cover a broad range of statistical methods topics such longitudinal data analysis, statistical genetics, graphical modeling, measurement error and machine learning. He has been heavily involved in genetic association and other epidemiological, studies of breast cancer, at Karolinska Institutet, and has published on genetic risk prediction models of breast cancer.