Lawrence Joseph
Bayesian statistics
Professor
E-mail: lawrence.joseph@mcgill.ca
Telephone: +1 (514) 934-1934 ext: 44713
FAX: (514) 934-8293
Complete CV
Division of Clinical Epidemiology
McGill University Health Centre
Royal Victoria Hospital
687 Pine Avenue West
V Building, Room V2.10
Montreal, Quebec
Canada, H3A 1A1

My global research interest is Bayesian biostatistical modeling. I am currently focussing on three particular topics:


Diagnostic tests in the absence of a gold standard It is common in population screening surveys or in the investigation of new diagnostic tests to have results from one or more tests investigating the same disease, but none of which can be considered perfectly error-free. Despite these limitations, it is important for clinical and public health practice to have the best possible estimates of disease prevalence and the properties of diagnostic tests, such as the sensitivity and specificity. Methodology for both design and analysis of diagnostic test data have been developed. Current research involves analyses for continuous tests and ROC curves.
Bayesian methods for study design and sample size determination Consideration of the optimal number of subjects and other design aspects is required in any study. Early research covered sample size methods for standard binomial, normal, and differences between two binomial or normal sampling situations. Current research investigates more complex univariate and multivariate situations, including design for diagnostic studies investigation of novel sample size criteria for prior robustness.
Biostatistical Consulting and Collaboration In addition to methodologic research, I also serve as the collaborating biostatistician on many projects in various areas of medical research. Among others, I am currently participating in projects investigating gastroenterological disorders, lupus nephritis, and cardiovascular diseases.
Software development New biostatistical methods cannot reach their full potential without the distribution of the software necessary for their implementation. User-friendly programs have been created for all of the above methodologies, and have been made freely available to other researchers through this web page.