<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=648880075207035&amp;ev=PageView&amp;noscript=1">

BayesiaLab Webinar Series

Bayesian Parameter Estimation for Individualized Drug Dosing

Wednesday, October 23, 2019, 11 a.m. – 12 p.m. (CDT, UTC-0500)

Abstract

Despite their similar names, Bayesian statistics and Bayesian networks are very different fields of study. While the former has been known and debated for well over two-hundred years, the latter only dates back to the mid-1980s.

Given the much longer exposure of scientists to Bayesian statistics, anything labeled "Bayesian" is often automatically associated with this field. However, that turns out to be a burden to Bayesian networks as Bayesian statistics have been — rightly or wrongly — subject to much criticism, particularly in the 20th century. Similarly, at first glance, BayesiaLab may be perceived as a tool for Bayesian statistics, not networks.

In this webinar, we first wish to explain what is most different between these two domains. Once the fundamental differences are established, we turn to the commonalities. As it turns out, almost as a by-product using of Bayesian network paradigm, Bayesian parameter estimation can be performed very conveniently with BayesiaLab. Perhaps one could even say that Bayesian parameter estimation can be more easily explained in the Bayesian network framework than on its "home turf" of Bayesian statistics.

Example: Individualized Drug Dosing for Coumarin Patients

Coumarin is a long-acting oral anticoagulant drug given to patients who suffer from thromboembolic conditions. Its dosage needs to be monitored and adjusted on a continuous basis for each patient. Given the long-acting nature of the drug, the response to a dosage change can only be observed with a significant delay, which proves challenging for establishing a steady long-term dosage level.

We use this dosage adjustment process to illustrate the Bayesian belief updating of hyperparameters within a Bayesian network model. The proposed approach shows how a Bayesian network can help in systematically reducing the uncertainty regarding an individual patient's dosage response with the objective of reaching the ideal therapeutic range more quickly.

Webinar Registration for Oct. 23, 11 a.m. (CDT, UTC-05)