Hybrid computational modeling demonstrates the utility of simulating complex cellular networks in type 1 diabetes.

Persistent destruction of pancreatic β-cells in type 1 diabetes (T1D) results from multifaceted pancreatic cellular interactions in various phase progressions. Owing to the inherent heterogeneity of coupled nonlinear systems, computational modeling based on T1D etiology help achieve a systematic understanding of biological processes and T1D health outcomes. The main challenge is to design such a reliable framework to analyze the highly orchestrated biology of T1D based on the knowledge of cellular networks and biological parameters. We constructed a novel hybrid in-silico computational model to unravel T1D onset, progression, and prevention in a non-obese-diabetic mouse model. The computational approach that integrates mathematical modeling, agent-based modeling, and advanced statistical methods allows for modeling key biological parameters and time-dependent spatial networks of cell behaviors. By integrating interactions between multiple cell types, model results captured the individual-specific dynamics of T1D progression and were validated against experimental data for the number of infiltrating CD8⁺T-cells. Our simulation results uncovered the correlation between five auto-destructive mechanisms identifying a combination of potential therapeutic strategies: the average lifespan of cytotoxic CD8⁺T-cells in islets; the initial number of apoptotic β-cells; recruitment rate of dendritic-cells (DCs); binding sites on DCs for naïve CD8⁺T-cells; and time required for DCs movement. Results from therapy-directed simulations further suggest the efficacy of proposed therapeutic strategies depends upon the type and time of administering therapy interventions and the administered amount of therapeutic dose. Our findings show modeling immunogenicity that underlies autoimmune T1D and identifying autoantigens that serve as potential biomarkers are two pressing parameters to predict disease onset and progression. Author summary: Pancreatic β-cells secreting insulin upon metabolic demand are the core to regulate glucose concentration for healthy states. Type 1 diabetes (T1D) is a group of metabolic disorders in which β-cells are targeted by biased decisions of the immune system. The barrier in sampling experimental data from pancreatic tissues in high-risk T1D subjects and the complexity of the mechanisms regulating T1D makes the use of quantitative modeling approaches an intriguing opportunity to analyze this disease. To decipher complex system behaviors resulting from inter- and intra-cellular and signaling networks linking the immune system and metabolism during T1D progression, we developed a hybrid computational framework to unravel high computational complexity levels for individual-specific T1D development in non-obese diabetic mice. We also discovered non-intuitive biological parameters, the potential for therapeutic strategies. These effective strategies associated with hybrid agent-based modeling can serve as a prototype for in- silico experiments of therapy-directed trials potentially regulating T1D progression. We expect that the proposed methods will play an important role in the quest for standardized applications of agent-based modeling to simulating complex disease progression. [ABSTRACT FROM AUTHOR]