Tools

The EMOD QuickStart introduces the basics of IDM's Epidemiological MODeling software (EMOD), which is a stochastic, agent-based model. The lessons in the QuickStart provide the fundamentals needed to understand the framework of the model and illustrate a variety of model capabilities. The lessons provided in the QuickStart are modular and can be completed in any order. 

After completing the QuickStart lessons, we recommend that you consult the EMOD documentation to learn more about advanced features of the model. IDM provides downloadable tutorial files to illustrate several different disease modeling scenarios.

IDM’s primary software tool, Epidemiological MODeling software (EMOD), simulates the spread of disease to help determine the combination of health policies and intervention strategies that can lead to disease eradication. EMOD is a stochastic, mechanistic, agent-based model that simulates the actions and interactions of individuals within geographic areas to understand the disease dynamics in a population over time.
 
EMOD software uses a "generic" transmission model that can be configured to run many different diseases, such as influenza or measles. This base functionality is then inherited by models for particular modes of transmission or individual diseases. EMOD has models for malaria, HIV, tuberculosis, and typhoid that explicitly contain logic that simulates how those diseases are transmitted, prevented, and treated.
 
This layered software architecture enables extensive testing and reuse of model functionality. IDM has created a suite of over 600 scenarios which run daily to ensure errors are not introduced into the current models. A vital part to this suite of tests is our “scientific feature testing,” which involves carefully validating the model output with the mathematical formulas and distributions specified for each disease.

To learn how to use EMOD for your research, we recommend completing the lessons in the online QuickStart course then walking through the tutorials and simulation examples provided in the documentation.

For modeling scenarios that do not require modeling of individual agents, IDM provides Compartmental Modeling Software (CMS). CMS enables you to construct a stochastic compartmental model using a variety of different solvers. This provides a simple way for you to define the compartments and the differential equations that govern the transition from one compartment to another, while still introducing stochasticity into the simulations.
 
Compartmental models have the benefit of being simpler to configure and faster to run, allowing you to quickly evaluate the potential outcomes of a variety of disease transmission scenarios.

As a stochastic model, EMOD can only produce useful insights after analyzing data from many different simulations. idmtools is a suite of Python tools that makes it easier to run thousands of simulations on a high-performance computing cluster. idmtools has extensive support for EMOD but can also be used with other disease models. idmtools has extensive functionality for data pre- and post-processing, creating input files, sweeping across a range of parameter values, calibrating the model, plotting output, and much more.

idmtools source code is not yet publicly available, but you can contact IDM to request access.

easyVA is a user-friendly web-based interface that enables clinicians to input verbal autopsy data. Additionally, algorithm-based applications can use easyVA for faster certification of causes of death. Cause of death data can be exported in CSV format. You must contact IDM for an account to access easyVA.

The Vis-Tools visualization toolkit creates layered, animated visualizations from disease simulation output. By linking temporal and geospatial simulation data, it promotes deeper insight into the complexity of disease dynamics.
 
Vis-Tools is a set of Python tools that can ingest data from any model, not just EMOD or CMS. The colors and visual elements used to represent different data are highly customizable. With Vis-Tools, you can create maps with layers of animated data that represent incidence, population density, importation through migration, and any other data you desire.

SpatioTemporal Analysis and Mapping in Python (STAMP) is a set of tools used to clean up spatial data sets, create raster files for shape data, and visualize the data from a raster file in PNG format. STAMP provides support for the Demographic Health Survey file format and allows you to use Python to directly query and retrieve spatial data in geoJSON or SHP format.

RSV Trends is a web dashboard that uses Google searches to estimate the number of children hospitalized in the United States for respiratory syncytial virus, or RSV. This methodology allows estimates to be made in near real-time to enable more effective control measures. Previous sources of RSV hospitalization data typically have delays ranging from weeks to years.

Computational Modeling Platform Service (COMPS) is a web-based user interface that facilitates research by providing access to high-performance computing environments. The dashboard allows you to monitor simulations that are running or queued and manage cluster resources using a job-priority system.

You can also create demographics, migration, and weather input files from the COMPS interface. The spatial and temporal weather data includes air temperature, relative humidity, and rainfall for many geographic regions across the globe using weather station readings and satellite data. COMPS provides a visualization of weather patterns overlaid on regional maps.

COMPS also provide visualization capabilities. One is powerful charting functionality to visualize the output channels for simulations. A chart can include output for a single simulation or for multiple simulations. Viewing multiple simulations in a single chart (multi-chart) provides a fast, flexible way to filter simulations to view only data of interest. Additionally, the spatial channel viewer visualizes geospatial data over time. You can choose from a rich set of base maps on which to overlay the output of the simulation and a large library of gradients to customize the colors used for the data.

COMPS computing resources are available only to IDM researchers and their collaborators. However, we will share any of the input files for the available countries. Contact us for more information.
 

A visual tooThe HIV Dashboard provides visualization of the EMOD network transmission model of HIV transmission in the Kingdom of eSwatini, explicitly calibrated to the age-specific and sex-specific scale-up of anti-retroviral therapy (ART) and voluntary medical male circumcision, to simulate historical and future HIV incidence and mortality trajectories.

Covasim is a stochastic agent-based simulator designed to be used for COVID-19 (novel coronavirus, SARS-CoV-2) epidemic analyses. These include projections of indicators such as numbers of infections and peak hospital demand. Covasim can also be used to explore the potential impact of different interventions, including social distancing, school closures, testing, contact tracing, and quarantine.

SynthPops is used construct synthetic networks of people that satisfy statistical properties of real-world populations (such as the age distribution, household size, etc.). SynthPops can create generic populations with different network characteristics, as well as synthetic populations that interact in different layers of a multilayer contact network. These synthetic populations can then be used with agent-based models like COVID-19 Agent-based Simulator (Covasim) to simulate epidemics.

The COVID Health Systems model is intended to make projections for the number of AAC and ICU hospital beds needed for COVID-19 patients and to estimate the patient waiting time for those beds. It provides a graphical UI using the SIMUL8 platform.

 

RAINIER is a statistical approach for fitting SEIR epidemic models to case and mortality data. We use the approach to create models of COVID-19 transmission in King County, WA.