Panel Members: John Braun, Rob Deardon, Charles Mayo
|John Braun, PhD
Deputy Director, CANSSI-INCAS
Professor and Head of Computer Science, Mathematics, Physics and Statistics
Adjunct Professor, Dept. of Stat and Act Sci,
Simon Fraser University & University of Western Ontario
Abstract: Feature Selection with Confidence Intervals
The LASSO and its relatives, SCAD and MCP, are powerful methods for doing variable selection in high dimensional regression. However, confidence intervals for the resulting regression coefficients are not always satisfactory. We review the literature and suggest a completely new variable selection procedure which is often competitive with the LASSO and which is amenable to confidence interval construction.
|Rob Deardon, B.Sc (HONS), M.Sc, PhD
Faculty of Veterinary Medicine and Department of Mathematics & Statistics,
University of Calgary
Abstract: Artificial intelligence in infectious disease surveillance
Ongoing improvements in artificial intelligence technology offer up tantalizing opportunities in many areas of application, including infectious disease surveillance and epidemiology. The opportunities arise from both improvements in modelling techniques (e.g., deep learning), and a vastly increased capacity for collecting data, especially through automated systems. A particular area of interest, for both public health and agricultural officials, is the development and implementation of automated disease surveillance systems. Such systems might be used, for example, identify the onset of the influenza season or the emergence of a new strain of the disease, or identify an outbreak of a disease on a particular farm using sensory data. The goals here are typically to provide for alarm systems which identify the outbreaks as early as possible, and/or improve upon the accuracy of human identification of such outbreaks. Here, we shall discuss how modern machine learning classification algorithms can be used to improve on current disease surveillance methods, how successful they appear to be, and what problems are still to be solved. This will be done within the context of a number of human and livestock diseases, including influenza and mastitis in cattle.
|Charles S. Mayo, PhD, FAAPM
Associate Professor, Radiation Oncology
University of Michigan
Ann Arbor, MI
Abstract: Leveraging Clinical Healthcare Informatics and Analytics to Bring AI into Clinical Practice
The combination of big data with machine learning (ML) and artificial intelligence (AI) algorithms promise to fundamentally change the nature of how we develop, test and apply clinical insights in healthcare. By creating a high volume big data analytics resource system to integrate “real world” clinical experience into a standardized ontological framework and then applying it to feed these algorithms we are creating a pathway in our clinic toward increased ability to apply observational data driving discovery and informing decision frameworks. Enabling this shift requires physicists to work on several efforts on several fronts including: clinical practice standardization, development of databases and ontologies, modeling and profiling using statistical and machine learning, and treatment team coordination. Using recent examples including use of an of AI toxicity model to refine DVH constraints, ML profiling factors affecting of treatment and imaging times we will highlight the several roles physicists must play to enable this translation of AI into clinical practice.
Jennifer (Jenn) Jakobi, PhD
Dean Pro Tem College of Graduate Studies
Professor, Health and Exercise Sciences
The University of British Columbia – Okanagan Campus
Panel Members: Wayne Beckham (RSO), Lesley Buckley (RSO), Yani Picard (CNSC), John Schreiner (RSO)
Moderator: Michelle Nielsen
Abstract: As part of their mandate, the Canadian Nuclear Safety Commission (CNSC) routinely conducts inspections of Class II licensed facilities. The goal of these inspections is to improve the radiation safety program at the facility, while ensuring compliance with regulations. Preparation for an inspection, particularly a comprehensive Type I audit, requires significant effort on the part of both the licensee and the inspection team. This preparation extends beyond the Radiation Safety Officer alone and necessarily involves members of the entire program. Thorough preparation prior to an inspection can alleviate concern and stress on the part of staff participating in the inspection and is key to ensuring an efficient review by the inspection team. The session will present strategies for preparing for a CNSC inspection, from the viewpoint of both the licensee and of the CNSC. Panel members will share some tested practices they have used to ensure a smooth inspection.