.png){kind=icon}
.png){kind=icon}
Return
A multi-network spatial error workflow for disentangling digital connectivity and geographic spillovers: evidence from sexually transmitted infections
.png){kind=icon}
PRE
AI
F
T
Z
M
J
G
J
T
W
DOI：10.1080/13658816.2026.2680021.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Capturing spatial interaction remains a challenge in GIScience and spatial epidemiology, yet many studies rely on a single neighborhood definition. We proposed a multi-network spatial error workflow that treats spatial weight matrices as theory-guided synthetic connectivity channels and estimates channel-specific dependence parameters within one likelihood. Using county-level U.S. gonorrhea, chlamydia, and HIV outcomes (2018–2019), we controlled for established sexually transmitted infection (STI) covariates and evaluated candidate spatial error models based on queen contiguity ( WQ), Twitter-derived mobility connectivity ( WP), and Facebook-derived socio-spatial connectivity ( WS). Model choice used a two-stage procedure that screened candidates in-sample and then evaluated shortlisted models out-of-sample via leave-one-state-out prediction and held-out residual autocorrelation diagnostics. The selected dependence structures were pathogen-specific: HIV supported a joint adjacency-plus-mobility specification ( WQ+WP), chlamydia was dominated by mobility ( WP), and gonorrhea was dominated by socio-spatial connectivity ( WS). These channel choices were associated with systematic re-ordering of fitted risk rankings, but the resulting shifts remained model-derived rather than externally validated indicators of real-world transmission importance. This study supports a reproducible workflow for comparing and selecting adjacency-, mobility-, and socio-spatially defined neighborhood structures in large-scale spatial modeling, with applications beyond STIs.
Keywords:
Multi-network
spatial error model
spatial weights
spatial epidemiology
social media
infectious diseases
Journal
IF:
5.1
Papers:
2.7K
Citations:
9.3K
Organization
Cited Papers