Prediction of the spread of African swine fever through pig and carcass movements in Thailand using a network analysis and diffusion model

Chaithep Poolkhet; Suwicha Kasemsuwan; Sukanya Thongratsakul; Nattachai Warrasuth; Nuttavadee Pamaranon; Suphachai Nuanualsuwan

doi:10.7717/peerj.15359

Prediction of the spread of African swine fever through pig and carcass movements in Thailand using a network analysis and diffusion model

PeerJ. 2023 May 9:11:e15359. doi: 10.7717/peerj.15359. eCollection 2023.

Authors

Chaithep Poolkhet¹, Suwicha Kasemsuwan¹, Sukanya Thongratsakul¹, Nattachai Warrasuth², Nuttavadee Pamaranon², Suphachai Nuanualsuwan^{3

4}

Affiliations

¹ Veterinary Public Health, Kasetsart University, Kamphaeng Saen, Nakhon Pathom, Thailand.
² Department of Livestock Development, Ministry of Agriculture and Cooperatives, Bangkok, Thailand.
³ Department of Veterinary Public Health, Chulalongkorn University, Bangkok, Thailand.
⁴ Center of Excellence for Food and Water Risk Analysis (FAWRA), Faculty of Veterinary Science, Chulalongkorn University, Bangkok, Thailand.

Abstract

Background: African swine fever (ASF) is a serious contagious viral disease of pigs that affects the pig industry. This study aimed to evaluate the possible African swine fever (ASF) distribution using network analysis and a diffusion model through live pig, carcass, and pig product movement data.

Material and methods: Empirical movement data from Thailand for the year 2019 were used, and expert opinions were sought to evaluate network properties and the diffusion model. The networks were presented as live pig movement and carcass movement data at the provincial and district levels. For network analysis, a descriptive network analysis was performed using outdegree, indegree, betweenness, fragmentation, and power law distribution, and cutpoints were used to describe movement patterns. For the diffusion model, we simulated each network using spatially different infected locations, patterns, and initial infection sites. Based on expert opinions, the initial infection site, the probability of ASF occurrence, and the probability of the initial infected adopter were selected for the appropriated network. In this study, we also simulated networks under varying network parameters to predict the infection speed.

Results and conclusions: The total number of movements recorded was 2,594,364. These were divided into 403,408 (403,408/2,594,364; 15.55%) for live pigs and 2,190,956 (2,190,956/2,594,364; 84.45%) for carcasses. We found that carcass movement at the provincial level showed the highest outdegree (mean = 342.554, standard deviation (SD) = 900.528) and indegree values (mean = 342.554, SD = 665.509). In addition, the outdegree and indegree presented similar mean values and the degree distributions of both district networks followed a power-law function. The network of live pigs at provincial level showed the highest value for betweenness (mean = 0.011, SD = 0.017), and the network of live pigs at provincial level showed the highest value for fragmentation (mean = 0.027, SD = 0.005). Our simulation data indicated that the disease occurred randomly due to live pig and carcass movements along the central and western regions of Thailand, causing the rapid spread of ASF. Without control measures, it could spread to all provinces within 5- and 3-time units and in all districts within 21- and 30-time units for the network of live pigs and carcasses, respectively. This study assists the authorities to plan control and preventive measures and limit economic losses caused by ASF.

Keywords: African swine fever (ASF); Carcasses and products; Diffusion model; Network analysis; Pig; Thailand.

Publication types

Evaluation Study
Research Support, Non-U.S. Gov't

MeSH terms

African Swine Fever Virus*
African Swine Fever* / epidemiology
Animal Husbandry
Animals
Industry
Swine
Thailand / epidemiology

Grants and funding

This work was supported by the Agricultural Research Development Agency (Public Organization) of Thailand (Grant No. PRP6105022740). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.