Intervention models
An intervention model represents a deliberate action intended to change epidemic transmission, disease progression, or health outcomes. The important modelling principle is not simply to “change a parameter”, but to identify which biological or behavioural mechanism the intervention actually changes.
Start from the epidemic model
For the simple SIR model,
\[\frac{dS}{dt}=-\beta\frac{SI}{N},\qquad \frac{dI}{dt}=\beta\frac{SI}{N}-\gamma I,\qquad \frac{dR}{dt}=\gamma I.\]The infection process is controlled by \(\beta\), while removal from infectiousness is controlled by \(\gamma\). An intervention can therefore act by changing transmission, changing infectious duration, moving people into another compartment, or changing later disease outcomes.
What does \(\beta\) represent?
It is useful to separate two ideas that are often combined inside \(\beta\). In a simple model we may write
\[\boxed{\beta=cq},\]| Symbol | Meaning |
|---|---|
| \(c\) | average effective contact rate: how often potentially infectious contacts occur |
| \(q\) | probability of transmission per effective contact |
| \(\beta=cq\) | effective transmission-rate parameter used in the compartment model |
This distinction helps show how different interventions act. A lockdown or distancing measure mainly reduces \(c\). Masks or other measures that reduce transmission during contact may reduce \(q\). Both can reduce \(\beta\), but through different mechanisms.
Transmission-reducing interventions
Contact reduction
Measures such as distancing, school closure, workplace closure, restrictions on gatherings, or lockdown can be represented by reducing the contact component of transmission.
\[c_1\longrightarrow c_2,\qquad c_2Reducing transmission per contact
Other interventions may mainly reduce the chance that an infectious contact causes transmission:
\[q_1\longrightarrow q_2,\qquad q_2Keeping \(c\) and \(q\) conceptually separate can make intervention assumptions easier to interpret than changing \(\beta\) without explanation.
Representing an intervention in time
If an intervention begins at time \(t_L\), transmission can be represented using one value before the intervention and a lower value afterwards:
Before \(t_L\), transmission follows the original regime. At \(t_L\), the intervention changes the modelled transmission process.
The lower value after \(t_L\) does not mean infections fall immediately. It means the rate at which new infections are generated is reduced. The infectious population may continue rising for some time if transmission is still sufficiently strong or because people infected before the intervention are still progressing through the disease process.
How does an intervention change epidemic growth?
For the simple SIR model,
\[R_{\mathrm{eff}}(t)=\frac{\beta(t)}{\gamma}\frac{S(t)}{N}.\]An intervention that lowers \(\beta(t)\) lowers \(R_{\mathrm{eff}}\). If it becomes smaller than 1, the infectious population tends to decline:
\[R_{\mathrm{eff}}<1.\]Other intervention mechanisms
| Intervention | Main model mechanism | Possible mathematical representation |
|---|---|---|
| Distancing / lockdown | fewer effective contacts | reduce \(c\), hence reduce \(\beta\) |
| Masking or reduced transmission during contact | lower probability of transmission per contact | reduce \(q\), hence reduce \(\beta\) |
| Testing and isolation | infectious people transmit for less time or have fewer contacts | increase effective removal rate or move \(I\) into an isolated class |
| Treatment | faster recovery or reduced severe progression | increase recovery rate or modify hospitalisation/death transitions |
| Vaccination | reduced susceptibility, infectiousness, or severe disease | add vaccinated compartments or modify infection/progression terms |
| Quarantine of exposed individuals | reduce transmission before or during infectiousness | add quarantine compartments or alter progression/contact terms |
Testing and isolation as a compartment intervention
Instead of changing \(\beta\) alone, testing and isolation can be represented explicitly by adding an isolated compartment \(Q\):
This can be more informative than simply lowering \(\beta\), because the model explicitly records how many infectious individuals are isolated and allows their transmission rate to differ from that of non-isolated infectious individuals.
Intervention timing, strength and duration
| Feature | Meaning in the model |
|---|---|
| Timing | when the intervention begins |
| Strength | how much the relevant parameter or transition rate changes |
| Duration | how long the altered regime remains in place |
Two interventions with the same strength can produce different outcomes if one begins earlier. Likewise, a strong but short intervention may delay transmission without preventing later resurgence if parameters return to their original values while many people remain susceptible.
What outcomes should intervention models compare?
The appropriate outcome depends on the decision being studied. Common model outputs include peak infectious prevalence, total infections, epidemic duration, hospital or critical-care demand, deaths, probability of exceeding capacity, and the time at which a threshold is crossed.