rflex_zones: Determine zones for flexibly shaped spatial scan test

Description

rflex_zones determines the unique zones to consider for the flexibly shaped spatial scan test of Tango and Takahashi (2012). The algorithm uses a breadth-first search to find all subgraphs connected to each vertex (region) in the data set of size \(k\) or less with the constraint that the middle p-value of each region must be less than alpha1.

Usage

rflex_zones(
  nn,
  w,
  cases,
  ex,
  alpha1 = 0.2,
  type = "poisson",
  pop = NULL,
  cl = NULL,
  loop = FALSE,
  verbose = FALSE,
  pfreq = 1
)

Arguments

An n by k matrix providing the k nearest neighbors of each region, presumably produced by the knn function.

A binary spatial adjacency matrix for the regions.

cases

The number of cases observed in each region.

The expected number of cases for each region. The default is calculated under the constant risk hypothesis.

alpha1

The middle p-value threshold.

type

The type of scan statistic to compute. The default is "poisson". The other choice is "binomial".

pop

The population size associated with each region. The default is NULL since this argument is only needed for type = "binomial".

A cluster object created by makeCluster, or an integer to indicate number of child-processes (integer values are ignored on Windows) for parallel evaluations (see Details on performance).

loop

A logical value indicating whether a loop should be used to implement the function instead of pbapply. The default is FALSE. If TRUE, then memory-saving steps are also taken.

verbose

A logical value indicating whether progress messages should be provided. The default is FALSE. If both loop and verbose are TRUE, informative messages are displayed that can be useful for diagnosing where the sequences of connected subgraphs are slowing down or having problems.

pfreq

The frequency that messages are reported from the loop (if verbose = TRUE). The default is pfreq = 1, meaning a message is returned for each index of the loop.

Value

Returns a list of zones to consider for clustering. Each element of the list contains a vector with the location ids of the regions in that zone.

References

Tango, T. and Takahashi, K. (2012), A flexible spatial scan statistic with a restricted likelihood ratio for detecting disease clusters. Statist. Med., 31: 4207-4218. <doi:10.1002/sim.5478>

Examples

Run this code

# NOT RUN {
data(nydf)
data(nyw)
coords = cbind(nydf$x, nydf$y)
nn = knn(coords, k = 5)
cases = floor(nydf$cases)
pop = nydf$pop
ex = pop * sum(cases)/sum(pop)
# zones for poisson model
pzones = rflex_zones(nn, w = nyw, cases = cases, ex = ex)
# }
# NOT RUN {
pzones = rflex_zones(nn, w = nyw, cases = cases,
                      ex = ex, verbose = TRUE)
# zones for binomial model
bzones = rflex_zones(nn, w = nyw, cases = cases, ex = ex,
                     type = "binomial", pop = pop)
# }