Predictive Modeling for Decision-making in Public Health (2022-2023)

Important note: This team is a joint project between Duke and Duke Kunshan University. This team seeks student participants from both campuses. Prospective applicants should be prepared to collaborate across countries, which may necessitate virtual meetings in the early morning or late evening to accommodate time zone differences.

Background

As countries across the globe grapple with limited public health budgets, policymakers are seeking to optimize the impact of public health expenses and investments. Choosing which interventions to deploy based on reliable predictions of cost and effectiveness can save countries money and increase access to care. Cost-effectiveness analysis (CEA) is a process that models the effects of interventions on a population, predicting their costs and benefits across time horizons and estimating a set of metrics for making informed decisions on the best resource allocation. 

When applying CEA to health interventions, uncertainty must be factored into the modeling. Disease progression varies among individuals, and the effects of interventions are also variable. Current tools for CEA can mask this complexity, adding a layer of modeling that hides the mathematical details of Markov chains (a system that transitions from one state to another according to probabilistic rules). However, a large set of results and tools for Markov chains exist, which could be transferred to public health practitioners to make conducting CEA a smoother and more rigorous process.

Project Description

This project team will advance cost-effectiveness analysis in public health by incorporating analysis techniques, abstraction languages and automation tools developed in mathematics and engineering for Markov models. Specifically, the team aims to match the needs of CEA with the capabilities of the stochastic reward nets (SRNs) modeling formalism. Team members will follow an incremental approach, whereby each increment produces original results from the interdisciplinary exchanges.

First, the team will investigate the definition and application of a modeling/analysis process for conducting CEA using the existing features of the SRN modeling formalism. Team members will share their public health and engineering expertise to define a common ground and terminology for further joint work. 

Next, the team will develop an abstraction layer on top of SRNs, specifically designed for supporting CEA, by conducting a substantial analysis of literature on CEA in order to gather a more complete understanding of its application. 

Finally, the team will build on the identified abstractions to define an entirely new modeling and analysis approach to support CEA. The team will develop a domain specific language (DSL) to provide public health practitioners with a friendly interface for conducting CEA using SRNs.

Anticipated Outputs

Papers for publication; technical documentation; materials for short course on application of SRNs to CEA

Student Opportunities

Ideally, this team will be comprised of 3 graduate students and 6 undergraduate students. The team will recruit students from Duke Kunshan University and Duke University for collaboration among two distinct research communities: computer engineers working on model-driven assessment of systems; and global health researchers dealing with CEA. Preferred skills and backgrounds include experience on predictive modeling, public health policy evaluation, data analysis and programming. Expertise on bibliography management tools is highly appreciated. 

A substantial amount of new knowledge must be acquired to work on the project, which pertains to advanced modeling and its multifaceted applications. Therefore, students should to have a genuine interest in multidisciplinary research, an open mindset and the willingness to learn new topics. 

Students will have the opportunity to complement their background knowledge with important competences on mathematical modeling, evidence-based decision-making and computational thinking. Team members will have the opportunity to present and disseminate project results, prepare manuscripts for publication and create teaching materials. Graduate students will have opportunities to gain leadership skills.

The project will hold weekly meetings. Subgroups will be formed to facilitate collaborative research, knowledge sharing and learning. Subgroups will be temporary, and will deal with short-term project objectives. Each subgroup will be led by a graduate student, and will be composed by undergraduate students and at least one faculty team member. Selected team members will be a part of a Steering Committee, jointly providing technical guidance and oversight of project activities. The Steering Committee will decide on project resource management, milestones and deadlines. Graduate students will supervise the work of undergraduate students and report to faculty leaders. 

Selected students will have the opportunity to travel to conference and scientific events for presenting project results. 

Ivan Mura will serve as project manager.

Timing

Summer 2022 – Summer 2023

  • Summer 2022 (optional): Collect and organize learning materials; begin literature and tools review
  • Fall 2022: Initiate first increment; attend lectures; analyze SRN case studies; prepare manuscript
  • Spring 2023: Initiate second increment; continue literature review and analysis; form subgroups; complete mansucript; receive DSLs training; implement prototype. 
  • Summer 2023 (optional): Complete prototype; work on technical documentation; write scientific papers; finalize instructional materials for short course

Crediting

Academic credit available for fall and spring semesters; summer funding available

 

Image: Duke Kunshan University in March 2016, by Chris Hildreth

Duke Kunshan University.

Team Leaders

  • Meifang Chen, Duke Kunshan University
  • Ivan Mura, Duke Kunshan University
  • Truls Ostbye, School of Medicine-Family Medicine and Community Health
  • Kishor Trivedi, Pratt School of Engineering-Electrical & Computer Engineering

Theme(s):