-
Notifications
You must be signed in to change notification settings - Fork 0
Contacts
The contacts system enables parameter values to be combined across populations. In Optima TB, this is currently only used for force-of-infection (FOI) to implement infection across populations.
The specification of interaction between populations occurs in two parts
- In the Cascade file, a parameter that combines values across populations is marked with the special rule
avg_contacts_in - In the databook, interactions weights are specified in a matrix in the 'Population Contacts' sheet
| Interaction Impact Weights | 0-2 | 3-14 | 15-64 | 65+ | 15-64 (HIV+) | 65+ (HIV+) |
|---|---|---|---|---|---|---|
| 0-2 | 5 | 1 | 1 | 3 | 1 | |
| 3-14 | 1 | 5 | 3 | 3 | 1 | |
| 15-64 | 1 | 3 | 5 | 1 | ||
| 65+ | 3 | 3 | 1 | 5 | 1 | 1 |
| 15-64 (HIV+) | 1 | 1 | 10 | 1 | ||
| 65+ (HIV+) | 1 | 1 | 1 | 5 |
This matrix represents interactions from each row, to each column.
The workflow for computing a parameter tagged by avg_contacts_in is as follows:
- First, the parameter value is computed as specified in the cascade within each population independently
- For each population, the parameter value is replaced by a weighted average across all populations
The weight factor is the product of the interaction weight in the contacts matrix, and the population size of the remote population. Because the weighted average is normalized by the sum of weights, the effect is that the parameter value lies within the range of values computed within each population separately, biased towards the population with the largest interaction_weight*size. For example, if the weights were all the same, then if a small population interacts with a large population, the FOI experienced by the small population will be close to the value in the large population.