Math Biology Seminar: Mike Irvine

  • Date: 10/04/2017
  • Time: 14:00
Mike Irvine, UBC

University of British Columbia


Linking mathematical models to public health policy: Use of Bayesian inference and Markov models in evaluating the current opioid overdose crisis in British Columbia.


The rapid increase of fentanyl and fentanyl analogues in British Columbia has led to a public health emergency being declared and a rapid increase in overdoses and overdose-related deaths in the province. Numerous interventions have been proposed in response, however it is not clear how to evaluate these interventions where the rate of overdoses is rapidly changing. We introduce a Poisson hidden Markov model to incorporate knowledge on ambulance-attended overdoses, fentanyl-related deaths and illicit-drug related deaths. We explicitly model the use of Take Home Naloxone kits (THN), an opioid agonist used in reversing an overdose that has been widely distributed. The model was fit using a Bayesian framework with informative priors, taking into account expert knowledge and literature-based rate estimates. We use the fitted model to estimate the total number of deaths averted due to the use of THN and explore a number of counterfactual scenarios including if THN was distributed sooner and if the size of the at-risk population was reduced.

Other Information: 

Location: ESB 4127