Near-term forecasting of companion animal tick paralysis incidence: An iterative ensemble model

Nicholas Clark, Tatiana Proboste, Guyan Weerasinghe and Ricardo Soares Magalhães

Tick-borne illnesses constitute a diverse group of debilitating conditions for pet dogs and cats around the world. In Australia, thousands of domestic dogs are admitted to emergency veterinary clinics due to tick paralysis each year. These admissions are highly seasonal and may be associated with changing environmental conditions, suggesting models that learn from environmental patterns to forecast the oncoming tick season could inform pet owners and clinicians about changing risks. In this paper we use a series of statistical forecasting models to analyse and predict tick paralysis admissions to veterinary clinics in a tick paralysis hotspot in Queensland, Australia. Our approach is novel in that we combine individual models into a superior ensemble that is trained to reduce forecast uncertainty, giving more accurate estimates of what the coming tick season will look like. Our model consistently outperforms a field-leading benchmark while uncovering important patterns about environmental drivers of paralysis tick exposure, including changes to levels of moist vegetation and maximum temperature. We also demonstrate how our model can be used to automatically produce forecasts of tick paralysis admissions as new data become available. This can have important implications for designing improved early warning systems for tick-borne illness.

Read the full article by clicking on the following link:

Clark, Nicholas, Proboste, Tatiana, Weerasinghe, Guyan & Soares Magalhães, Ricardo. (2022). Near-term forecasting of companion animal tick paralysis incidence: An iterative ensemble model. PLoS computational biology. 18. e1009874. 10.1371/journal.pcbi.1009874. 


Ixodes holocyclus Australian Paralysis Tick Catching The Eye Flikr 41012546034 53e8d16668 c

Ixodes holocyclus, Australian Paralysis Tick by "Catching The Eye" on Flikr (41012546034_53e8d16668_c)