Inferring spreading dynamics of diseases and in neural networks

by Jonas Christoph Dehning
Doctoral thesis
Date of Examination:2024-11-15
Date of issue:2025-06-11
Advisor:Prof. Dr. Viola Priesemann
Referee:Prof. Dr. Viola Priesemann
Referee:Prof. Dr. Stefan Klumpp
Persistent Address: http://dx.doi.org/10.53846/goediss-11280

 

 

Files in this item

Name:dehning_phd_thesis.pdf
Size:6.53Mb
Format:PDF

The following license files are associated with this item:

Abstract

English

The rapid advancement in information technology over the past few decades has provided ample computational power and access to diverse data sources. In conjunction with recent inference methods, we can now examine the intricate dynamics of complex systems in more detail than ever before. Two complex systems that have been proven particularly challenging to study are the spreading of neural activity in the brain and infectious diseases in society. The complexity of cortical neural dynamics arises from the difficulty in inferring global characteristics from the limited, subsampled, and localized measurements that can be obtained simultaneously. In contrast, the complexity of the spread of infectious diseases comes from the dynamic nature of human behavior, which is shaped by individual social environments. In this thesis, I present three distinct projects that address these challenges: (1) We investigated the impact of a stimulus on the primary somatosensory cortex in mice and its subsequent effects on the secondary somatosensory cortex, including the decrease of stimulus detection probability due to recurrent dynamics. (2) Using the European championship during the COVID-19 pandemic as a case study, we inferred the influence of simultaneous social gatherings on disease transmission. (3) We explored how spreading dynamics inform the optimal allocation of vaccines under conditions of limited availability. For the systematic analysis of such spreading dynamics, I developed a comprehensive toolbox for inferring parameters of differential equations, incorporating Bayesian inference methods. Complementing the projects presented in this thesis, I contributed to a dozen other research endeavors during my PhD, which are not extensively discussed here. Looking ahead, these projects demonstrate that further advancements in inference techniques will unlock new research opportunities, enabling the integration of diverse data sources and improving our ability to quantify and manage the inherent uncertainty associated with dynamic systems.
Keywords: Complex systems; Bayesian inference; Statistics; Differential equations
 
Statistik