Module 52: Using Stochastic Simulation to Inform the Design of Emergency Intervention Trials in Low- and Middle-Income Countries

Research Areas:


Building on the ideas for Module 48: A review of mapping techniques of slums using machine learning, remote sensing, and volunteered geographic information, and delving more into the the Big Data aspects of our IGERT, this module develops models to inform the design of emergency medical care in low and middle income countries. Trainee Vania Wang worked with a team led by Dr. Carl Pearson, a physicist who develops applied models for public health. They submitted an abstract containing the information below to the Consortium of Universities for Global Health Conference.


Emergency intervention trials are difficult to implement because of ethical and logistical challenges. Such trials are further complicated in low- and middle-income countries (LMIC); these settings compound typical trial challenges with the added complexity of constrained resources, limited infrastructure, and more heterogeneous demography and geography.

However, these additional complexities can be managed with more detailed analysis.  One way to include that detail is via agent-based, stochastic simulation. We describe a framework to apply this approach to evaluate emergency intervention trial designs, divided into four distinct phases: event population generation, randomization according to trial protocol[s], simulation of treatment and control outcomes, and evaluation of the protocol analyses.  We demonstrate this framework applied to a real-world emergency medical intervention trial in an LMIC setting.


Using our framework, we developed a trial simulation based on a proposed protocol for a multisite, randomized controlled trial in South Africa. We generate events based on a census of prehospital emergency service responses, randomize according to multiple designs, simulate multiple intervention mechanisms, and evaluate a variety of outcomes.


Using our framework, we are to distinguish between a variety of scenarios within a trial in a comprehensible way and estimate important characteristics of the proposed protocols, including power and sample size, false positive rates, safety signals, and the effects of different operational approaches to the trial.


As medical and public health organizations seek to expand operations in LMICs, they may need to use new approaches to meet the scientific and ethical standards of trials traditionally carried out in resource-rich settings.  The increasingly global availability of low-cost computational resources offers one approach to meeting this challenge.


Active Quarters:

  • Fall 2018: Vania Wang