Spatial epidemiology: ZIP Gibbs, kernel relative risk, Oden Ipop, prospective scan, ecological regression, wombling

ZipGibbs: Bayesian zero-inflated Poisson relative risks by data augmentation with Marsaglia-Tsang gammas on Philox streams. KernelRelativeRisk: Kelsall-Diggle Gaussian-kernel log relative risk. OdenIpop: Oden's population-adjusted Moran statistic with a multinomial Monte Carlo test. ProspectiveScan: Kulldorff's prospective space-time scan (Poisson, windows ending at the last time). PoissonEcological: IRLS Poisson regression with offset and quasi-Poisson scale. ArealWombling: boundary likelihood values on adjacent pairs and barrier edges. BufferRateRatio: exposure-zone rate ratio with a conditional binomial test. NeedBasedAllocation: weighted capitation. FunnelControlLimits: Spiegelhalter funnel limits. GhoseDrugFilter: the Ghose drug-likeness rule. Indices are 0-based, as the Python arm.

Usage

ZipGibbs(y, expected, n_iter, a = 1, b = 1, burn = 0, seed = 0)

KernelRelativeRisk(cases, controls, points, bandwidth)

OdenIpop(cases, population, W, nsim = 0, seed = 0)

ProspectiveScan(
  counts,
  expected,
  coords,
  max_window = 3,
  max_pop_frac = 0.5,
  nsim = 99,
  seed = 0
)

PoissonEcological(y, expected, X)

ArealWombling(values, A, quantile = 0.8)

BufferRateRatio(cases, population, coords, sources, radius)

NeedBasedAllocation(budget, population, need_index, floor_share = 0)

FunnelControlLimits(
  target,
  sizes,
  z = c(1.95996398454005, 3.09023230616781),
  kind = "proportion",
  phi = 1
)

GhoseDrugFilter(mw, logp, mr, n_atoms)

Arguments

y, cases: Case counts.
expected: Expected counts.
n_iter, burn: Gibbs sweeps and burn-in.
a, b: Gamma prior shape and rate.
seed: Philox seed.
controls: Control point coordinates.
points: Evaluation points.
bandwidth: Kernel bandwidth.
population: Populations.
W: Spatial weights (diagonal allowed).
nsim: Monte Carlo replicates.
counts: Times by regions case counts.
coords: Region centroids.
max_window: Longest time window.
max_pop_frac: Largest share of the expected total in a zone.
X: Covariates (an intercept is added).
values: Area values.
A: Adjacency matrix.
quantile: Barrier quantile.
sources: Point-source coordinates.
radius: Buffer radius.
budget: Total budget.
need_index: Need indices.
floor_share: Share of the budget split equally.
target: Target proportion.
sizes: Denominators or expected counts.
z: Normal quantiles of the limits.
kind: "proportion" or "smr".
phi: Overdispersion factor.
mw, logp, mr, n_atoms: Molecular weight, logP, molar refractivity and atom count.

Value

A list, vector or logical vector.

References

Lambert, D. (1992). Technometrics 34, 1-14. Kelsall, J. E. and Diggle, P. J. (1995). Statistics in Medicine 14, 2335-2342. Oden, N. (1995). Statistics in Medicine 14, 17-26. Kulldorff, M. (2001). JRSS A 164, 61-72. Lu, H. and Carlin, B. P. (2005). Geographical Analysis 37, 265-285. Spiegelhalter, D. J. (2005). Statistics in Medicine 24, 1185-1202. Ghose, A. K. et al. (1999). J. Comb. Chem. 1, 55-68.

Examples

PoissonEcological(c(2, 4, 8), c(1, 1, 1), matrix(0:2))$beta
#> [1] 0.6931472 0.6931472
FunnelControlLimits(0.1, 100, z = 2)$upper
#> [[1]]
#> [1] 0.16
#>