Agent-based modeling of the interaction between CD8⁺ T cells and Beta cells in type 1 diabetes

We propose an agent-based model for the simulation of the autoimmune response in T1D. The model incorporates cell behavior from various rules derived from the current literature and is implemented on a high-performance computing system, which enables the simulation of a significant portion of the islets in the mouse pancreas. Simulation results indicate that the model is able to capture the trends that emerge during the progression of the autoimmunity. The multi-scale nature of the model enables definition of rules or equations that govern cellular or sub-cellular level phenomena and observation of the outcomes at the tissue scale. It is expected that such a model would facilitate in vivo clinical studies through rapid testing of hypotheses and planning of future experiments by providing insight into disease progression at different scales, some of which may not be obtained easily in clinical studies. Furthermore, the modular structure of the model simplifies tasks such as the addition of new cell types, and the definition or modification of different behaviors of the environment and the cells with ease.本研究提出一种用于1型糖尿病（Type 1 Diabetes, T1D）自身免疫反应模拟的基于智能体的模型（agent-based model）。该模型整合了基于现有文献推导得到的多种细胞行为规则，并基于高性能计算系统实现，可完成小鼠胰腺中大量胰岛的模拟任务。模拟结果表明，该模型能够精准捕捉自身免疫反应进展过程中出现的动态趋势。该模型的多尺度特性支持定义调控细胞或亚细胞层面现象的规则与方程，并可在组织尺度上观测模拟结果。预期此类模型可通过快速验证假说、规划后续实验，为不同尺度下的疾病进展机制研究提供见解，其中部分尺度的信息在临床研究中难以直接获取，从而助力体内（in vivo）临床研究。此外，该模型的模块化结构可简化各类操作，例如轻松新增细胞类型，以及灵活定义或修改环境与细胞的各类行为模式。