LEMMAABMv4 is an agent-based model of COVID-19 transmission dynamics, currently parameterized to represent the city and county of San Francisco. Future development may focus on generalizing data importation and processing for use in other settings. Transmission proceeds through a modified Susceptible-Exposed-Infected-Recovered (SEIR) model that allows for inclusion of pre- and asymptomatic states ($I_p$ and $I_a$ , respectively) and age-dependent variability in symptom severity ($Im$, mildly symptomatic, and $I{mh}$ to $I_h$ for symptomatic pre-hospitalization and hospitalized, respectively) and mortality. Agent movement between census tracts is informed by Safegraph mobility metrics derived from anonymized mobile phone panels. There are no explicit network connections, rather all agents are members of households, workplaces, and communities which they attend depending on circumstances of the simulation and transmission occurs according to a location-specific FOI for each location at each time step.
devtools::install_github("cmhoove14/LEMMAABMv4")
Installation will load all scripts necessary to set up and simulate the model, but the data folder also contains pre-processed data inputs that will allow users to simulate the model right out of the box. Forking the repository may therefore be the best route to quick and easy setup and simulation.
The main simulation function is covid_abm_v4 which takes inputs:
bta_base - baseline transmission probability bta_hh - multiplier on household transmission probability bta_work - multiplier on workplace transmission probability bta_sip_red - multiplier on transmission probability following onset of first shelter in place to account for microbehaviors affecting transmission that are not explicitly included in the model data_inputs - all data inputs including synthetic agents, safegraph movement and stay at home metrics, etc. see documentation input_pars - list of lists with additional model parameters and setup vax_phases - list of lists delineating vaccine phases visitors - logical of whether to model visitors into target population (requires visitors setup in data_inputs) testing - logical of whether to model testing adaptive - logical of whether to model adaptive testing vaccination - logical of whether to model vaccination verbose - logical of whether to return progress summaries at each time step store_extra - logical of whether to store extra output files from simulation The bash folder also contains scripts to run in parallel via bash scripts in which model inputs are passed in shell