Substantiation of statistical model to describe and predict risks of tick bites for population

Abstract

Incidence of tick-borne encephalitis and other tick-borne infections correlates with a number of people applying for medical aid due to tick bites. Obviously, the number of registered tick bites is proportionate to people’s economic and recreational activities on an endemic territory and the quantity of hungry ticks. In its turn, the quantity of ticks depends on abundance of main hosts for blood-feeding stages but with a certain time lag caused by their life cycle parameters such as molting to the next stage, diapauses, and apparent seasonality in a continental boreal climate zone. Our research goal was to analyze and synthesize an adequate formalized/parameterized statistical model to describe and predict risks of tick bites for population. To describe dynamics and to predict a number of people bitten by ticks exemplified by the Sverdlovsk region, we used several linear (by parameters) logistic regression models. We applied a multimodel inference framework to assess whether the observed dynamics was described adequately. Long-tern dynamics of the number of people bitten by ticks in the Sverdlovsk region is characterized with an occurring high-amplitude slow long-wave oscillation (circadecadal one, with a quasi-period being approximately 10 years) and a short-wave 2–3-year cyclicity. The former may be associated with climatic rhythm and socioeconomic trends; the latter may be caused by biotic factors. By using the logit-regression model, we showed that the number of small mammals, both in the previous year and at the beginning of the current tick activity season can be a valuable predictor of a risk for population to be bitten by ticks. Predictive values of the created statistical model adequately describe an initial time series of chances/probabilities of tick bites.