The ICARUS ISI demonstrator’s scope is set within the context of the global spreading of epidemics (like Covid-19) in environments characterized by many degrees of complexity. The developed modelling tools in ISI Foundation aim at better understanding various phenomena related to the spreading of infectious diseases, which depend on multiple factors (e.g. disease characteristic features, topology of the interaction network, etc.) and at supporting policy makers in case of public health emergencies like the ongoing Covid-19 pandemic crisis. Thanks to this approach, it is possible to estimate the unfolding of an epidemic event and to evaluate the efficacy of various intervention strategies and control measures to be undertaken in order to contain or mitigate the effects of the disease outbreak.
The Objective of the ISI Demonstrator in ICARUS
The main objective of the ISI Demonstrator scenario in ICARUS is to implement non-incremental improvements to the current models by integrating additional available aviation-related data, like the travellers’ age structure. The process requires to characterize statistically the datasets and to model the real-world processes that produced the data: the novel modelling capabilities shall be analysed in terms of accuracy of the predictions, both in historical and current epidemic events. Using data stratified with the additional information will allow to improve the accuracy of the modelling approach and study indicators not previously accessible in the simulated scenarios.
While nowadays it is accepted that airline mobility is the key driver of the international spread of most infectious diseases affecting the human population, at the same time travel restrictions and advisories due to epidemic threats lead to important decline in air traffic, seriously influencing the airline industry. The current Covid-19 pandemic crisis has highlighted these considerations, and we decided to redirect part of our ICARUS activities towards the study of the ongoing epidemic threat.
More specifically, the demonstrator scenario aims at assessing the dependence of the attack rate (overall number of infections in a population) and its public health, demographic and economic impact on population demographics linked to airline mobility data. Census information for the various sub-populations shall be matched with age-stratified aviation data about passengers, refining the development of containment strategies and intervention measures.
Preparation
In the first phase of the ICARUS project, we spent some effort trying to access detailed data about passenger traffic from airline distribution systems, that would allow us to run the fundamental Demonstrator scenario dealing with age-stratification but also to explore the possibility of extending it with additional information about return bookings and tickets, as a proxy for average length of stay. This approach would have allowed testing alternative disease spreading modelling approaches exploring the time scale separation and effective interaction between short-distance commuting and long-distance travel modes.
We then devoted our activities towards exploring and collecting various datasets originating from national and international statistical offices worldwide, together with acquiring commercial aviation passenger data. One crucial step of the Demonstrator aims at constructing an age-stratified network for the long-range mobility of individuals through airline transportation, which is a key component for modelling the spreading of infectious diseases worldwide.
In order to perform more accurate and realistic simulations of infectious disease spreading, the computational epidemiological model is being extended by implementing a demographic structure for the population of each census area. To this purpose, 16 age brackets have been identified, defined in such a way that the results of the simulations can be accurately remapped to the five age groups proposed by the World Health Organization (WHO), thus allowing for an easier comparison between the output results of the model with actual epidemic data. With respect to the long-range mobility represented by air travels, this stratification of the population into 16 age groups also requires matching distributions of passengers per age bracket.
The Development of the ISI Disease Spreading Demonstrator
Since the evolution of an infectious disease outbreak strongly depends on contact patterns among individuals, the Demonstrator development focused on model improvements taking into account this feature during the epidemic dynamics. The contacts among individuals are represented by means of age-specific 16 × 16 contact matrices, computed from the synthetic populations built from observational data. In this case, the basic reproduction number R0, which represents the number of secondary cases generated by a typical index case in a fully susceptible population, becomes dependent on the contact matrices too. The use of contact matrices implies a more complex expression for the so-called force-of-infection, which gives the transition probability of any susceptible individual due to effective contact with infectious people.
The usage of the ICARUS infrastructure, due to the peculiar scientific context of the Demonstrator, was originally intended in relation to the opportunity of exploring and exploiting airline passenger data to enhance the understanding of global and local disease spreading phenomena. Given the extremely careful and demanding work required in order to combine and harmonize the datasets needed and used for building a long-range mobility network, the Demonstrator decided to exploit the powerful functionalities provided by the ICARUS platform for the various steps of data preparation, cleaning, imputation and pre-processing.
This integrated approach helps the automatization of such a procedure and the data quality and integrity checks, allowing to work on setting up a pipeline for the continuous update of the fundamental data sources used in the epidemiological model. Following this approach, additional test cases have been identified and added to the ISI Demonstrator scenario and shall be explained in a next blog post.
Blog post prepared by ISI.
