Using pet insurance claims to predict occurrence of vector-borne and zoonotic disease in humans

Abstract

AbstractTaking a One Health approach to infectious diseases common to both dogs and people, pet insurance claims from 2008-2022 in the United States were compared to publicly available CDC-based data on human cases for Lyme disease, giardia, and Valley Fever (coccidioidomycosis). Despite having very different causative agents and etiologies, the disease trends for these three diseases were very similar between people and dogs both geographically and temporally. We furthermore demonstrated that adding dog data to the human data improves prediction models for those same diseases. With machine learning prediction tools for the pet insurance to increase prediction times and alert public health officials, pet insurance data could be a helpful tool to predict and detect diseases by estimating even earlier the effects of these common exposure diseases on human health. We also show the spatiotemporal distribution of intestinal worm diagnoses in dogs, and while it could not be directly compared to human data because the corresponding disease in humans (soil-transmitted helminths) has not been well monitored recently. However, these data can help inform researchers and public health workers.