Summary
Computational advancements and the development of powerful agent-based simulation frameworks have allowed us to model unprecedented realism in mosquito-transmitted diseases – a threat that millions of people face every year. While this enhanced realism has already provided important insights, agent-based models are still often outperformed in disease forecasting by their idealized mathematical counterparts. This might be caused by a lack in our fundamental understanding of how spatial dynamics need to be appropriately modeled. To date, in depth sensitivity analyses of complex agent-based models that would help to address this issue are hardly conducted, because these complex models are very parameter rich. Thus, the required model recalibration and comparisons are very runtime intensive. I introduce Gaussian processes as a powerful statistical framework that will significantly advance sensitivity analysis for agent-based models and allow us to assess the relative input importance of different model parameters in unprecedented detail. The proposed research could further deepen our understanding of how spatial structure affects host-pathogen dynamics and be a valuable contribution to the field of sensitivity analysis for agent-based models. I expect to find that spatial structure changes disease transmission potential profoundly and can cause spatial-specific model dynamics. I hope that my findings contribute to the improvement of disease forecasting and evaluation of proposed intervention strategies.
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/101025586 |
| Start date: | 01-12-2021 |
| End date: | 30-11-2024 |
| Total budget - Public funding: | 252 349,44 Euro - 252 349,00 Euro |
Cordis data
Original description
Computational advancements and the development of powerful agent-based simulation frameworks have allowed us to model unprecedented realism in mosquito-transmitted diseases – a threat that millions of people face every year. While this enhanced realism has already provided important insights, agent-based models are still often outperformed in disease forecasting by their idealized mathematical counterparts. This might be caused by a lack in our fundamental understanding of how spatial dynamics need to be appropriately modeled. To date, in depth sensitivity analyses of complex agent-based models that would help to address this issue are hardly conducted, because these complex models are very parameter rich. Thus, the required model recalibration and comparisons are very runtime intensive. I introduce Gaussian processes as a powerful statistical framework that will significantly advance sensitivity analysis for agent-based models and allow us to assess the relative input importance of different model parameters in unprecedented detail. The proposed research could further deepen our understanding of how spatial structure affects host-pathogen dynamics and be a valuable contribution to the field of sensitivity analysis for agent-based models. I expect to find that spatial structure changes disease transmission potential profoundly and can cause spatial-specific model dynamics. I hope that my findings contribute to the improvement of disease forecasting and evaluation of proposed intervention strategies.Status
TERMINATEDCall topic
MSCA-IF-2020Update Date
28-04-2024
Images
No images available.
Geographical location(s)
Structured mapping
