Pro-Active Routing for Emergency Testing in Pandemics

In the COVID-19 pandemic, we saw that rapid and efficient testing can effectively slow the spread of a virus. However, especially at the beginning of the pandemic, testing was extraordinarily time- consuming and expensive, and testing resources were not available everywhere on the scale required. Authorities responded to this challenge with innovative ideas: in Vienna, for example, a bicycle-based team of mobile testers was created, which could be requested in case of suspicion. The mobile tester fleet could then collect samples from suspected cases, bring them for evaluation, and thus help reduce further spread. Based on the idea of a mobile fleet of testers, we now want to investigate in this research project how scarce resources can be used sensibly in a dynamic environment by planning ahead. For this purpose, modern methods and tools from agent-based simulation, data analysis and dynamic vehicle routing are used. For the COVID-19 example, this means, for example, not waiting for potential suspicious cases, but using simulation to predict how a pandemic will spread, so that suitable tests can then be carried out in advance. This first requires a forecast by means of a spread model, which is already available for the COVID-19 example due to the extensive data situation. But at which point should one test already in advance? The underlying problem must be mathematically formulated and analyzed for this purpose. In addition to agent-based simulation, innovative learning methods are also used to ensure that the test resources are used as effectively as possible. This then results in a catalog of policies that are examined in the simulation for their effectiveness. For what kind of problems is such an approach useful at all? Which policies of predictive planning work particularly well? Abstracting from the COVID-19 example, we also want to analyze similar problems. When is the interaction of predictive planning in combination with machine learning helpful? How can detailed simulations improve forecasts? And how does one computationally feasibly integrate information from highly complex simulations into predictive planning? Examples here extend far beyond COVID- 19, e.g., to demand forecasting for delivery services. Participants in the project are Univ.-Prof. Jan Fabian Ehmke from the University of Vienna, who will focus on demand modeling. In addition, there is the expertise of Univ.-Prof. Dr. Marlin Ulmer, University of Magdeburg, who will be responsible for learning procedures related to predictive vehicle routing. Evaluation and demand generation will be done by Dr. Niki Popper (TU Wien), who is an expert in the area of agent-based simulation.

We live in a world full of dynamic and heterogeneous socio-economical and socio-ecological challenges. In the project "Pro-Active Routing for Emergency Testing in Pandemics", we addressed these challenges with simulation-based analytics (SimBA). From a methodology perspective, SimBA entailed data-driven, smart decision making including causal and expert information. The starting point was data on pandemic dispersal, which could be derived from large databases. In addition, simulation of socio-economical relationships and systems provided synthetic data on pandemic spread. Our simulation focused on microscopic, agent-based models. The data and information created were transformed into information models, which described complex socio-economical relationships compactly. This information was considered in anticipatory decision models, which allow for making smart resource allocation decisions. The impact of these decisions was evaluated by simulation models again. In particular, in several papers and in a joint review paper, we focused on descriptive, predictive and prescriptive analytics for resource allocation in pandemic management. As the COVID-19 pandemic demonstrated, having the right containment resources, such as testing kits, vaccines, and personal protective equipment like masks, at the right time and location to achieve effective pandemic containment and to mitigate pandemic spread is both critical and challenging. High-quality predictions are essential for allocating such scarce resources, and substantial work has been conducted on descriptive and predictive analytics during the COVID-19 pandemic. Policymakers attempted to incorporate the resulting information into immediate decision making in reaction to the predicted spread. However, to be prepared for the next pandemic, we worked on prescriptive analytics models that account for the dynamic interaction between predicted pandemic development and prescribed resource allocation. This integrated view has been limited in the literature, and we tried to contribute with this project by closing this gap. To this end, we provided several papers on analytics research conducted on the case of the COVID-19 pandemic and its impact on resource allocation. We developed dynamic and stochastic resource allocation models that consider historical data of pandemics spread and optimized resource allocation as needed. Data from simulation and the COVID-19 pandemic were included in extended optimization models and can help to assign the right containment resources where needed in the future.