Bayesian Networks for Health Economics and Public Policy Research
This seminar was held on June 6, 2018, in Toronto. A video recording of this particular seminar is currently not available. However, you can access all the presentation materials below. Also, please note that we hosted a very similar event in New York City later in the year, for which a recording is available.
In this seminar, we will illustrate five prototypical examples of applied research with Bayesian networks and BayesiaLab related to health economics and public policy:
- Diagnostic decision support: using Bayesian networks for cost-effective evidence-seeking in diagnosing coronary heart disease.
- Quantifying the value of information in field triage for defining adaptive trauma activation thresholds for the optimal utilization of hospital resources.
- Developing universal policies under extreme uncertainty, e.g., in the absence of epidemiological data: "test & treat" vs. presumptive malaria treatment in sub-Saharan Africa.
- The Chicago Condom Campaign: Simpson's paradox rears its ugly head.
- Causal inference from observational healthcare data: using machine learning and the Disjunctive Cause Criterion to reduce—not eliminate—the need for causal assumptions from domain experts.
Each case study will include a theoretical presentation and a practical demonstration using the BayesiaLab software platform.
- Presentation Slides (PDF, 58.4 MB)
- Example 1: Diagnostic Decision Support
- Example 2: Trauma Activation Policy
- Example 3: Optimizing Health Policies Under Uncertainty
- BayesiaLab Network File (XBL, 2 KB)
- Example 4: Chicago Condom Campaign
- Example 5:
Upcoming Seminars, Webinars, and Conferences
|Live Webinar||June 26, 2019||11:00 – 12:00 (CDT, UTC-5)||Causal Counterfactuals for Contribution Analysis — Explaining a Misunderstood Concept with Bayesian Networks|
|Live Webinar||July 10, 2019||11:00 – 12:00 (CDT, UTC-5)||Black Swans & Bayesian Networks — Jointly Representing Common and Rare Events|
|Please check out our archive of recordings of previous events.|