2016/02/10 by Guy Harling, Harling, Guy, Jukka‐Pekka Onnela +1
Mathematics · Physics and Astronomy · #COVID-19 epidemiological studies #Complex Network Analysis Techniques #FOS: Computer and information sciences #FOS: Physical sciences #Other Statistics (stat.OT) #Physics and Society (physics.soc-ph) #Social and Information Networks (cs.SI)
paper · pdf · doi:10.48550/arxiv.1602.03434
openalex publication_date 2016/02/10 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
Understanding how person-to-person contagious processes spread through a\npopulation requires accurate information on connections between population\nmembers. However, such connectivity data, when collected via interview, is\noften incomplete due to partial recall, respondent fatigue or study design,≠.g., fixed choice designs (FCD) truncate out-degree by limiting the number of\ncontacts each respondent can report. Past research has shown how FCD truncation\naffects network properties, but its implications for predicted speed and size\nof spreading processes remain largely unexplored. To study the impact of degree\ntruncation on spreading processes, we generated collections of synthetic\nnetworks containing specific properties (degree distribution,\ndegree-assortativity, clustering), and also used empirical social network data\nfrom 75 villages in Karnataka, India. We simulated FCD using various truncation\nthresholds and ran a susceptible-infectious-recovered (SIR) process on each\nnetwork. We found that spreading processes propagated on truncated networks\nresulted in slower and smaller epidemics, with a sudden decrease in prediction\naccuracy at a level of truncation that varied by network type. Our results have\nimplications beyond FCD to truncation due to any limited sampling from a larger\nnetwork. We conclude that knowledge of network structure is important for\nunderstanding the accuracy of predictions of process spread on degree truncated\nnetworks.\n