What has happened in the world over the past year and a half has brought dramatically to general attention the subject of infectious disease spreading and the study of pandemics evolution and mitigation strategies.
The GLEAM framework
ISI Foundation has been active for many years on the field of epidemiological research and on the investigation of large-scale disease spreading phenomena in particular. One of the main outcomes of the scientific activity has been the development of a flexible and realistic computational infrastructure, built upon real world data, that is capable of simulating the worldwide spreading of various kinds of infectious pathogens, also allowing to compare baseline scenarios with multiple intervention strategies and containment measures. The Global Epidemic and Mobility model (GLEAM) is a stochastic metapopulation modelling framework that allows to simulate the spatiotemporal spreading of different infectious diseases at the global level, considering over 3200 subpopulations in about 230 different countries and territories around the world, and the mobility connections between them.
A crucial element for being able to implement a model capable of realistic predictions and analyses, together with the development of a robust mathematical design that captures the complexity of the system, is the availability of quality data to use as constitutional input. GLEAM embeds core datasets about worldwide population density, flight networks, daily commuting flows, and intra-population contact patterns.
The role of ICARUS
The main interest in participating to the ICARUS project for the ISI Foundation is related to the integration of additional aviation data to improve the capabilities and the accuracy of the epidemiological modelling framework.
As already mentioned, airline traffic data is a key component for modelling and simulating long-range human mobility. Various characteristics of aviation passenger data can be valuable for epidemiological modelling; for the execution of the Demonstrator scenario, the focus was on one of the most relevant, passengers’ age distribution, which wasn’t available before on global scale data-driven models.
Using the ICARUS platform, researchers from ISI can search for interesting datasets, and define appropriate queries to trigger when new data become available. The platform also allows to pre-process the datasets with multiple data preparation functions, and to run aggregation algorithms that help in generating the mobility networks that are used by the computational framework for performing the disease spreading simulations.
Ideally the target was to acquire average distributions of passenger fluxes per age groups along all the major air routes worldwide, aggregated monthly or yearly. Since those data could not be accessed due to legal bindings preventing them to be distributed, multiple alternative data sources were collected, originating from passenger surveys and statistical aggregations performed by tourism offices, institutional agencies and private companies. This information, once properly characterised and rearranged, had been integrated with the origin-destination bookings dataset sold by OAG, which is used by the GLEAM mobility component.
The execution of the demonstrator scenario involved various activities: starting from the collection and the analysis of the available datasets, their processing and standardization in order to use them in the computational infrastructure, the rewriting of the equations and algorithms of the theoretical model, the refactoring of the simulation code, its validation and tuning, and finally the validation of the new models with real case scenarios, in order to assess the quality of the improvements.
The benefits and key take-aways
The novel epidemiological models, embedding age stratification and coupling intra-population age structure with age distributions about passenger mobility, allowed ISI to compute previously not available output data and comparisons. The upgraded framework was tested and validated both using historical data series about seasonal flu and with a global-scale pandemic scenario; COVID-19 outbreak attracted a lot of resources and various analyses about the ongoing emergency have been performed using the age-stratified approach. The improved modelling infrastructure showed a measurable increase in terms of accuracy of the predictions, highlighting even more the importance of quality input data with higher level of detail.
The ICARUS platform has all the chances to become one of the top aggregators for aviation data sources, and ISI Foundation as well as other research institutions will benefit from its functionalities even if not belonging to the aviation sector.
Blogpost prepared by ISI
Photo by National Cancer Institute on Unsplash
