Heterogeneous biting risk¶
This page provides an overview of how and when to use the EMOD features for heterogeneous biting risk.
What is heterogeneous biting risk and why would you want to include it?¶
Many studies have identified substantial variance in the number of bites that individuals within a population experience, which may be attributable to a combination of factors, such as individuals’ age, physiology, behavior, and location (Irvine et al 2018, Cooper et al 2019). Including biting risk heterogeneity in malaria transmission models is important for accurately capturing prevalence, incidence, probability of elimination, and other important simulation outcomes. To capture this heterogeneity in EMOD, we specify both age-specific risks and individual relative risks, which can increase or decrease the number of bites an individual tends to experience relative to the mean risk in the population.
Changing the distribution of relative biting risks does not change the total number of bites given to the population, but it does change how those bites are distributed among individuals. People with higher risk have a higher probability of receiving bites than people with lower risk. Because it is a relative biting rate, the exact values assigned do not matter, only their relation to the sum of all risks in the population. For example, giving everyone a value of 10 is the same as giving everyone a value of 1.
A brief description of the different ways of specifying heterogeneous biting in EMOD is given below, along with links to the main documentation pages for each type.
Age_Dependent_Biting_Risk_Type¶
Where is it specified?¶
The relative biting risk among people of different ages is specified in the configuration file with the Age_Dependent_Biting_Risk_Type parameter. There are three options for the type of distribution used (‘OFF’, ‘LINEAR’, or ‘SURFACE_AREA_DEPENDENT’, which are explained in Infectivity and transmission.
Who does it apply to?¶
If Age_Dependent_Biting_Risk_Type is enabled, the biting risk of everyone in the population is modified according to their age. This includes people who were created at the beginning of the simulation as well as people who are born during the simulation.
How does serialization affect it?¶
Yes, one can change this parameter when reading from a serialized file. The age dependent part of an individual’s biting risk is calculated each time step.
When should it be used?¶
If you aim to capture more realistic relative biting risks across ages in your simulation, it is generally a good idea to enable age-dependent biting risk.
Biting risk demographics¶
Where is it specified?¶
The distribution of relative biting risks across the entire population is specified in the demographics file using the RiskDistributionFlag (which provides the type of distribution used) along with RiskDistribution1 and RiskDistribution2 parameters (which specify the shape of the distribution). Details on the values for these parameters can be found in Demographics parameters.
Note
It is necessary to also set Enable_Demographics_Risk to 1 in the configuration file, otherwise the demographics risks will be ignored in the simulation. You must also set Enable_Demographics_Birth to 1 if you want newborns to get different relative biting risks.
Who does it apply to?¶
When the relative biting risk distribution is set in the demographics file, all individuals that are created at initialization and all individuals that are born during the simulation have relative biting risks that are drawn from that distribution.
How does serialization affect it?¶
An individual that is deserialized from the file will have whatever relative biting risk they had when they were deserialized. However, if Enable_Demographics_Risk and Enable_Demographics_Birth were serialized on, then newborns will get new relative biting risk values based on the RiskDistributionFlag, RiskDistribution1, and RiskDistribution2 parameters in the demographics file.
Individual’s risks do not change at serialization.
When should it be used?¶
The demographics approach for specifying relative biting risk is well-suited for simulations where you want to include a more realistic distribution of risks across all individuals in the population.
Biting risk intervention¶
Where is it specified?¶
The biting risk intervention campaign is a campaign that changes the biting risk of a specified group of people at a particular point in time in the simulation. Details on setting up biting risk intervention campaigns can be found in BitingRisk.
Who does it apply to?¶
Like other intervention campaigns, individuals with certain individual properties, ages, etc. can be targeted, or it can be distributed to the entire population. Only people who exist in the simulation at the time the campaign occurs are affected. This contrasts with the demographics approach to setting the risk distribution, where all individuals are initialized or born with risks drawn from the distribution.
Will an individual’s current risk be overwritten?¶
Implementing a biting risk intervention campaign overwrites the existing relative biting risk values of targeted individuals (regardless of these biting risks were set using the demographics file or a previous biting risk intervention). The new biting risk values that are assigned are independent of the biting risk a person had previously.
How does serialization affect it?¶
Individual’s risks do not change at serialization. However, whenever there is a new BitingRisk intervention, it will overwrite existing relative risk values.
When should it be used?¶
The intervention version of setting biting risks is best suited for situations where you want a subset of the population to have a different risk than everyone else. For example, if you want a group of people who work in high-risk areas to have substantially higher risk than the general population, you can use IPs and biting risk campaigns to target those individuals.
Common questions¶
- Do heterogeneous biting risks apply when the model is configured to use forced EIR instead of mechanistic mosquito bites?
No, but there are different parameters that can be used to specify age-based heterogeneity.