The early identification of the spreading patterns of an epidemic infectious disease is an important first step towards the adoption of effective interventions. We developed a simple regression-based method to estimate the directional speed of a disease's spread, which can be easily applied with a limited dataset. We tested the method using simulation tools, then applied it on a real case study of an African Swine Fever (ASF) outbreak identified in late 2021 in northwestern Italy. Simulations showed that, when carcass detection rates were <0.1, the model produced negatively biased estimates of the ASF-affected area, with the average bias being about -10%. When detection rates were >0.1, the model produced asymptotically unbiased and progressively more predictable estimates. The model produced rather different estimates of ASF's spreading speed in different directions of northern Italy, with the average speed ranging from 33 to 90 m/day. The resulting ASF-infected areas of the outbreak were estimated to be 2216 km2, about 80% bigger than the ones identified only thorough field-collected carcasses. Additionally, we estimated that the actual initial date of the ASF outbreak was 145 days earlier than the day of first notification. We recommend the use of this or similar inferential tools as a quick, initial way to assess an epidemic's patterns in its early stages and inform quick and timely management actions.
Keywords: Asfaviridae; Sus scrofa; carcasses; disease surveillance; early detection; regression models; wild boar.