Adding heterogeneity
One of the benefits of an agent-based model like EMOD over compartmental models is that the model can be configured to capture heterogeneity in population demographics, migration patterns, disease transmissibility, climate, interventions, and more. This heterogeneity can affect the overall course of the disease outbreak and campaign interventions aimed at controlling it.
Demographics
Built-in demographics options are available for running EMOD simulations, or you can create customized demographics files to represent particular locations. It is generally recommended that you create a demographics file instead of using built-in demographics.
Every individual within the simulation has a variety of attributes, represented by continuous or discrete state variables. Some are static throughout life, and others dynamically change through the course of the simulation, through response either to aging or to simulation events (such as infection). Static attributes are assigned upon instantiation (simulation initialization or birth after the beginning of the simulation) and include gender, time of birth, time of non-disease death, etc. Dynamic attributes include disease state, history of interventions, and more.
Vital dynamics
Vital dynamics within EMOD are derived from fertility and mortality tables that are passed to the model as input. Input demographic data can be used to construct a cumulative probability distribution function (CDF) of death date based on individuals' birth dates. Then, in the model, individual agents will be sampled stochastically from this CDF using an inverse transform of this distribution. Female agents similarly sample the age at next childbirth, if any, upon instantiation and birth of a previous child. Pregnancy is not linked to relationship status, although newly born individuals are linked to a mother. The fertility rate changes by simulation year and female age, and the range for available estimates depends on input data. Values outside of this range can be chosen by "clamping," or choosing the nearest value within the range. Clamping was also used when necessary to determine the non-disease mortality rate, which varies by gender, age, and simulation year.
For more information on the demographics file, see Demographics file.
Individual and node properties
One of the most powerful and flexible features of EMOD is the ability to assign properties to nodes or individuals that can then be used to target interventions or move individuals through a health care system. For example, you might assign various degrees of risk, socioeconomic status, intervention status, and more. In the generic, environmental, typhoid, airborne, and TBHIV simulation types, these properties can be leveraged to add heterogeneity in transmission based on the property values assigned to each individual. For example, you might configure higher transmission among school-age children.
Innate immune variation
Individuals in a population differ in their innate immune responses to malaria. EMOD can model this heterogeneity by drawing a per-individual innate immunity modifier (v) from a distribution configured in the demographics file (InnateImmuneDistributionFlag/InnateImmuneDistribution1/InnateImmuneDistribution2). The modifier persists for the individual's lifetime.
The Innate_Immune_Variation_Type configuration parameter controls how the drawn value is interpreted. In the formulas below, v is the drawn value, Pyrogenic_Threshold and Fever_IRBC_Kill_Rate are configuration parameters, and age is the individual's age in years.
NONE (default)
No innate immune variation. The distribution is ignored; all individuals use the configured Pyrogenic_Threshold and Fever_IRBC_Kill_Rate directly.
PYROGENIC_THRESHOLD
The drawn value scales the individual's pyrogenic threshold — the IRBC/µL density at which fever is triggered:
Higher values mean the individual tolerates higher parasite loads before becoming febrile. Bounded by Pyrogenic_Threshold_Min and Pyrogenic_Threshold_Max.
CYTOKINE_KILLING
The drawn value scales the individual's cytokine-mediated parasite killing rate:
Higher values produce a stronger innate killing response.
PYROGENIC_THRESHOLD_VS_AGE_CONCAVE
The pyrogenic threshold starts at v × Pyrogenic_Threshold and changes with age, reflecting
acquired tolerance. The threshold is recalculated every 3 months:
base = v × Pyrogenic_Threshold
if age < 2 years:
threshold = base + 0.035 × base × age
else:
threshold = base × 0.965 × exp(-0.09 × (age - 2)) + base × 0.1
For young children, the threshold increases linearly. After age 2, it decays exponentially toward 10% of the base value (the asymptotic tolerance level for adults). Bounded by Pyrogenic_Threshold_Min and Pyrogenic_Threshold_Max.
PYROGENIC_THRESHOLD_VS_AGE_INCREASING_AND_CYTOKINE_KILLING_INVERSE
Both the pyrogenic threshold and cytokine killing rate are modified. The distribution should be Uniform(0, 1) because the cytokine formula assumes v is in that range.
Pyrogenic threshold increases with age:
Capped at Pyrogenic_Threshold_Max. Once the cap is reached, the threshold remains fixed.
Cytokine killing rate varies inversely with v:
Individuals with a higher pyrogenic threshold (more tolerant of fever) have a lower cytokine killing rate, and vice versa.
Related parameters
| Parameter | Description |
|---|---|
| Pyrogenic_Threshold | Base IRBC/µL level at which innate inflammatory response is half-maximal (default: 1000) |
| Pyrogenic_Threshold_Min | Floor for calculated pyrogenic threshold (default: 0.1) |
| Pyrogenic_Threshold_Max | Ceiling for calculated pyrogenic threshold (default: 100000) |
| Fever_IRBC_Kill_Rate | Maximum IRBC kill rate from innate inflammatory response (default: 0.15) |
See Immunity configuration for the full parameter list.
To configure innate immune variation in emodpy-malaria, use MalariaDemographics.set_innate_immune_distribution(). This method automatically sets Innate_Immune_Variation_Type at task build time.
For more details on the innate immune response model, see Infection and immunity.
Transmission
In EMOD, transmission can only happen within a geographic node, and the population is "well-mixed" in each node. Heterogeneous transmission is modeled through biologically mechanistic parameters that control aspects of the simulation such as parasite density, symptom severity, mosquito bites, and more.
See Infectivity configuration parameters for more information on configuring transmission in this simulation type. Because HINT cannot be used with this simulation type, the parameter Enable_Heterogeneous_Intranode_Transmission in the configuration file must be set to 0 (zero).
Migration
EMOD can also simulate human and vector migration, which can be important in the transmission of many diseases. You can assign different characteristics to each geographic node to control how the disease spreads.
When Enable_Migration_Heterogeneity is enabled, each individual receives a personal migration rate multiplier drawn from a distribution configured in the demographics (MigrationHeterogeneityDistributionFlag/Distribution1/Distribution2). This multiplier persists for the individual's lifetime, creating realistic variation where some individuals are frequent travelers and others rarely leave their home node. This heterogeneity applies only to human migration; vector migration rates are controlled separately through habitat, food, and stay-put modifiers.
For more information, see Geographic migration.
For more information on how you can target campaign interventions to individuals or locations based on certain criteria, see model-campaign.