Utilization of spatially resolved data sources for an established agent-based model of Germany and its impact on predicted SARS-CoV-2 dynamics

• Applicants: Professor Dr.-Ing. Bernd Hellingrath; Professor Dr. André Karch
• Subject Area: Epidemiology and Medical Biometry/Statistics ; Data Management, Data-Intensive Systems, Computer Science Methods in Business Informatics
• Term: from 2021 to 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 492390948

The goal of this project is to integrate real-time spatial health-, mobility- and behavioural data in a previously developed agent-based simulation platform to provide reliable regional forecasts of age-specific incidence rates for a period of 2-4 weeks at any stage of an epidemic. The model will be based on the agent-based simulation platform EPIPREDICT. The system already offers a comprehensive population model of the German population (approx. 80 million agents) on federal state, district and municipality levels. However, apart from the population structure, the model does currently not include any spatial information about simulated agents. While the platform has been assessed for its general usability in examining local infection dynamics and intervention strategies retrospectively, it has never been intended to provide regional short-term forecasts. These generally require a higher temporal and spatial resolution of input data. Due to its high spatial resolution the EPIPREDICT population model offers the opportunity to close this gap by integrating real-time spatial health-, mobility- and behavioural- data. With the present project proposal, we plan to extend the platform to include this perspective.For this purpose, four types of regional real-time data will be considered in the simulation, enabling regional short-term forecasts: the current pandemic situation, current mobility, current contact- and preventive behaviour, and current locally enforced non-pharmaceutical interventions (NPIs). Our three main project objectives are: (1) the development of a spatial agent-based forecasting model, (2) the development of a modelling workflow enabling efficient regular forecasts and (3) the development of a dashboard to make simulation results publicly available.The work program is divided into the two project areas. First, the "Data Management" project area led by the department of Epidemiology (André Karch) concerns the regular compilation, management, analysis, and preparation of data on the current infection dynamics to be integrated in the model. Second, the "Development" project area led by the department of Information Systems (Bernd Hellingrath) focusses on the model- and method development, dashboard- and interface development, as well as the execution and evaluation of forecasts. To achieve our goals regarding the processing of spatial data, the project team will be advised by the Institute for Geoinformatics of the University of Münster (Christian Kray) who takes a supporting role.Although we intend the model to be used in the context of the current pandemic, our findings and the prototype are applicable to support future pandemics and containment efforts. The modelling workflow proposed here can serve as a feasibility study for the development of a nationwide regional early warning system, which could be implemented in a follow-up project.

• DFG Programme: Research Grants