Department of Electronic and Computer Engineering - Department of Electronic and Computer Engineering - Seminar
Seminar I : Dominating hyperplane regularization for sparse group Lasso regression
Speaker : Prof. Zeny Feng, Department of Mathematics and Statistics, University of Guelph
Hepatitis C health policy decision-making on prevention and treatment is complex. Preventive interventions such as syringe distribution programs can prevent forward transmission. Screening coupled with treatment is effective in reducing disease burden for hepatitis C virus (HCV) infections, which are asymptomatic until later stages of the disease. Drugs currently approved for HCV treatment are effective but costly. For HCV, where benefits of screening and treatment occur decades into the future, computational modeling is the only practical option for policy makers to analyze the impact of prevention and treatment.
In this talk, I present three studies that we had used computational models to generate evidence to inform HCV-related health policies: 1) A state-transition model that evaluate the cost-effectiveness of HCV-related screening and treatment interventions; 2) A back-calculation model based on a Bayesian MCMC algorithm to estimate chronic hepatitis C prevalence and undiagnosed proportion; and 3) An agent-based model that track hepatitis C elimination in an ongoing pandemic era.
Dominating hyperplane regularization for sparse group Lasso regression
Speaker : Prof. Zeny Feng, Department of Mathematics and Statistics, University of Guelph
Compositional data, measured by taxa counts at specified taxonomic rank, are prevalent in many biological fields including ecology and microbiology. In ecology, samples of benthic macroinvertebrate took from different aquatic sites were classified into taxonomic rank in an order of phylum, class, order, family, genus and species. At a given rank, the counts of species at each of the D taxa conditional on the total counts can be modeled by the Dirichlet multinomial (DM) distribution, which can accommodate multinomial over-dispersion. The model fitting in the presence of covariates can be challenging because the DM distribution falls outside the exponential family and the number of parameters is as high as pxD, where p is the number of covariates. With these challenges, we propose a sparse group LASSO in the regularized DM regression. We formulate an MM-algorithm using dominating hyperplane and Jensen’s inequalities to optimize the penalized DM regression likelihood. We apply the proposed method to analysis the association between water variables and benthic macroinvertebrate compositional data collected from Oil Sand Region in Canada.
Seminar II :
Using computational models to inform health policies in hepatitis C
Speaker : Prof. William W.L. Wong, CIHR Applied Public Health Chair, Associate Professor, School of Pharmacy, University of Waterloo
Hepatitis C health policy decision-making on prevention and treatment is complex. Preventive interventions such as syringe distribution programs can prevent forward transmission. Screening coupled with treatment is effective in reducing disease burden for hepatitis C virus (HCV) infections, which are asymptomatic until later stages of the disease. Drugs currently approved for HCV treatment are effective but costly. For HCV, where benefits of screening and treatment occur decades into the future, computational modeling is the only practical option for policy makers to analyze the impact of prevention and treatment.
In this talk, I present three studies that we had used computational models to generate evidence to inform HCV-related health policies: 1) A state-transition model that evaluate the cost-effectiveness of HCV-related screening and treatment interventions; 2) A back-calculation model based on a Bayesian MCMC algorithm to estimate chronic hepatitis C prevalence and undiagnosed proportion; and 3) An agent-based model that track hepatitis C elimination in an ongoing pandemic era.
Prof. Feng is a professor in statistics in the Department of Mathematics and Statistics, University of Guelph, Ontario, Canada. She is also a core faculty member in the interdisciplinary graduate programs of Bioinformatics, Collaborative Specialized Artificial Intelligence, and Data Science. Her research interests are on statistical genetics, statistical bioinformatics, environmental studies, and infectious disease modelling.
Dr. William W.L. Wong, is an Associate Professor in the School of Pharmacy at the University of Waterloo. He completed his PhD degree in Computer Science (Bioinformatics) at the University of Waterloo, and a Postdoctoral Fellowship in Medical Decision Making at the University of Toronto. He recently appointed by the Canadian Institutes of Health Research as an Applied Public Health Chair. He is also a member of the Ontario Health Technology Advisory committee (OHTAC) for Ontario Health, where he served as a computational modeling expert for the province. The committee reviews health technology assessments and makes recommendations on which health care services and devices should be publicly funded.
Dr. Wong’s research is focused on infectious diseases modeling, health services and outcomes research, particular in hepatitis B and C. His health services and outcomes research interests include quality of life research, costing and return on investment. Methodology research interests include advanced computational modeling techniques for health technology assessment, cost-effectiveness analysis, and pharmacoeconomics evaluation studies.