EpiModel

An open-source software platform for epidemic modeling in R.

  • Epidemics built on the TERGM statistical framework
  • Built-in SI / SIS / SIR models for exploration and teaching
  • Extendable API for research-level modeling

https://epimodel.org/

Jenness SM, Goodreau SM, Morris M. EpiModel: An R Package for Mathematical Modeling of Infectious Disease over Networks. Journal of Statistical Software. 2018;84(8):1-47.

NIH R01AI138783: EpiModel 2.0 (PI: Jenness).

Research Applications of EpiModel Across Diseases

Research Applications of EpiModel Across Diseases

https://github.com/EpiModel/EpiModel/wiki

EpiModel: Built-in and Extended

  • EpiModel is designed for both built-in (“toy models”) and user-defined extensions (“research models”).
  • This course focuses on built-in network models.
  • The User Defined branches above (disease types and process modules) are where research models live, and NME-II teaches you to write them.

COVID University DCM with EpiModel

  • A compartmental model for COVID on a university campus, led by Ben Lopman and Carol Liu, with Adrien Le Guillou.
  • Projects the impact of testing & quarantine and screening & isolation strategies.
  • Programmed and simulated in EpiModel.

COVID University DCM with EpiModel

https://epimodel.shinyapps.io/covid-university/

Network Model for MRSA

  • A network model of MRSA infection within a NICU setting.
  • Networks defined as shared hospital-worker contacts between infants.

Network Model for Seal Measles

Built with NME-II: a custom SEIR module on a dynamic network (Module 9)

EpiModel’s Modular Framework

  • Add any process of interest into the infectious-disease system, and use the base EpiModel tools (estimation, simulation, analysis, plotting).
    • These are tools that we are invested in helping you master!
  • It enforces modular thinking: building a complex system from small, interconnected building blocks.
  • This facilitates efficient expansion once you have a starting codebase.

HIV Preexposure Prophylaxis (PrEP)

  • Antiretroviral medication for prevention, provided to HIV-uninfected persons.
  • Decreases the biological risk of infection when an HIV-infected partner has uncontrolled viral replication.
  • Men who have sex with men (MSM) in the US are a high-priority population for PrEP.
  • 5% to 50% of MSM with indications currently use it.

HIV PrEP Indications as a Network Problem

US CDC PrEP Indications

US PHS/CDC guidelines indicated PrEP for those at “substantial risk” (2014, revised 2017 and 2021). For MSM, indications were:

  • Condomless anal intercourse (AI) in a monogamous partnership with a partner not recently tested for HIV
  • Condomless AI outside a monogamous partnership
  • AI (even with condoms) in a known serodiscordant partnership
  • Any non-HIV STI diagnosis

Clinicians screen for these over the past 6 months and reevaluate risk every 12 months.

Jenness SM, et al. Impact of CDC’s HIV Pre-Exposure Prophylaxis Guidelines among MSM in the United States. Journal of Infectious Diseases. 2016;214(12):1800-1807.

Partner Notification Interventions Across Networks

  • Direct patient delivery of antibiotic medications to sexual partners of diagnosed “index patients” (expedited partner therapy, EPT).
  • An example of contact-driven prevention related to partner notification (contact tracing).
  • Required historical network data on partnerships to represent the “look-back” period for identifying recent partners.
  • Epidemic model of HIV + NG + CT co-infection.
  • Counterfactual models explored different deployments of EPT by partnership type.

Built with NME-II: cumulative partnership history + custom modules (Module 9)

COVID’s Shock to the Sexual Network

  • Modeling a “shock” to the network from COVID-related sexual distancing, differential by partner type.
  • Gradual resumption of sexual activity over 2020.
  • Balancing decreased transmission from distancing against increased transmission from clinical service disruption.

Jenness SM, et al. Projected HIV and Bacterial STI Incidence Following COVID-Related Sexual Distancing and Clinical Service Interruption. Journal of Infectious Diseases. 2021;223(6):1019-28.

Built with NME-II: a full network HIV/STI model (Module 15)

PrEP Questions Network Models Can Answer

Evaluating CDC Guidelines

Risk Compensation & Adherence

Impact on Bacterial STIs

PrEP & Racial Disparities

Long-Acting Injectable PrEP

Empirical Data ⇝ Network Model Parameters

The ARTnet Study of MSM in the US (R21 MH112449): 4904 MSM reporting on 16,198 sexual partnerships.

  • Data-driven statistical models embedded within transmission models where primary data are available:
    • TERGMs for network structure → simulate
    • Poisson models for coital frequency → predict
    • Logit models for condom use → predict
  • Supports confounding adjustment and handling of parameter covariance and interactions.
  • Secondary data for more universal parameters (PrEP/ART effectiveness, per-act transmission probability).

pubmed.ncbi.nlm.nih.gov/32004795

Built with NME-II: data-driven network parameterization (Module 12)

What NME-II Teaches You to Build

The network features research models need, and where you build each in NME-II:

  • Flexible configurations and intuitive counterfactuals on network structure · Module 9
  • Data-driven parameterization from sampled egocentric data · Module 12
  • Multiple contact layers, each with its own formation and dissolution · Module 13
  • Open populations with demographic churn · already yours from NME-I (Module 7)
  • Ongoing temporal feedback between exogenous processes and structure · Module 14
  • A sparse network representation (networkLite) that scales up simulation · Module 15

And then, disease transmission models on top of it all.