Epidemiological Surveillance (spatial) Flashcards
(31 cards)
Pathogens move between different geographical units because of
movements of animals or humans or other hosts.
Spatial patterns arise from
- heterogeneity in the landscape
- large and dense cities v rural transmission
Spatial dispersal approximated by
exp(-d/a)
What determines the spatial spread of plant infectious diseases?
- contiguous spatial kernel
Triphragmium ulmariae
- rust fungus (Sphaerophragmiaceae) - meadowsweet rust gall
- chemically induces swelling on the lower surface of Filipendula ulmaria leaves
- implicated in survival
dots represent
- infection
- how many are there?
distance
(√(x2–x1)2+(y2–y1)2)
calculating force of spatial invasion
distances between infected and non-infected locations
Force of infection at any given location is defined by
- how close infected locations are to non-infected locations
- by increasing a the force of infection also increases.
compare how a spatial model estimating dispersal compares to a aspatial nullmodel
- create second model: infection risk uniform across all locations
- compare using anova
- outcome = infection status
- which is the better predictor via logistic regression: foi v nullmod?
when can you use logistic regression?
binary outcomes
likelihood of becoming infected
connectivity of an uninfected plant to infected plants
nullmod
No information about connectivity between uninfected and infected plants
Calculate log likelihood for Gaussian kernel and compare both models using AIC and
visualise their kernels.
AIC is a function of
- Model complexity
- Likelihood (how well the model reproduces the data)
better model according to AIC is the one that
explains the greatest amount of variation in the outcome using the fewest possible predictors
AIC =
- 2(number of parameters)- 2log(Likelihood)
- lower = better
human mobility patterns
not strictly determined by distance.
Models approximating mobility data usually take into account
- population of origin and destination location
- distance or travel time between them
- other variables that may influence travel patterns: attractiveness of a population (high degree of shopping/work opportunities)
Gravity model
movement volume between two communities depends inversely on distance, but bilinearly on size
Gravity model explanation
- assumes number of individuals travelling per unit time proportional to some power the source and destination populations
- decays with distance
- reflects transport infrastructure between locations
Assumptions of SIR:
- susceptible and infected individuals mix at random
- infectiousness does not change during course of infection in an individual
- no latent period
Spatial interaction matrix
relative connectivity (G)
Estimating the spatial interaction matrix
- Population per location
- Distances between them
