Spatio-temporal Predictive Risk Modelling of Virulent Newcastle Disease (vND) Cases Clusters in Zimbabwe from 1995-2014
Keywords:
geographic information systems, spatio-temporal risk modelling, hotspots, virulent Newcastle disease, clustersAbstract
Zimbabwe’s economic growth through the agriculture sector is periodically affected by outbreaks of
virulent Newcastle disease (vND). Newcastle disease is caused by avian paramyxovirus serotype 1
(APMV–1) virus of the Paramyxoviridae family, hence it is referred to as virulent Newcastle disease.
The virus is highly contagious, and is transmitted through faecal droppings, dead birds, direct contact
with infected birds, migration of infected birds, and other excretions. The symptoms of vND include
nervous system breakdown, respiratory distress, and reduced egg quality and production. Although
prior work has mapped outbreak locations, operational disease control requires risk-oriented
evidence that can support predictive surveillance and anticipatory intervention targeting. This study,
therefore, implements a spatio-temporal risk-modelling framework to identify districts and seasonal
windows with statistically significant elevated vND risk in Zimbabwe at 64 district level. The study
used spatial cluster and hotspot techniques — Kulldorff’s spatial scan statistic, and Getis Ord Gi*
test statistic, respectively. Getis-Ord Gi* statistics were used to detect hotspots of vND at the district
level, while the Kulldorff’s spatial scan statistic was used to detect spatial and space-time elevatedrisk
clusters of Newcastle cases. The results revealed that the western, eastern and northern districts
are characterised by hotspots of vND. From the study, we can conclude that the highest-likelihood
(primary) risk clusters of vND were detected in the north, east and northeastern districts of Zimbabwe.
This analysis also revealed that vND cases commonly occur between June and December, with June,
September and November having the highest incidences. These statistically derived risk signals
provide a defensible basis for predictive surveillance planning, enabling risk-based vaccination timing,
targeted biosecurity intensification, and prioritised monitoring in recurrent high-risk districts. While the
analysis provides a historical risk baseline, routine updating with contemporary surveillance data is
required before extrapolating risk patterns to current outbreak dynamics utilising techniques such as
machine learning and artificial intelligence.