eMB: Mouth to Mind: Leveraging network dynamical systems and software to understand diet, diabetes, dementia and modifiable risk factors delaying Alzheimer's disease

美国人正在输掉这场思想之战。肥胖和2型糖尿病加速了致命的、与年龄相关的神经退行性疾病的进展，如阿尔茨海默病（AD）。事实上，到2050年，AD将影响每个美国家庭; 65岁以上的美国人中有七分之一将患有AD，并且目前还没有明确有效的药物治疗方案。 在缺乏实用的AD药物治疗策略的情况下，了解AD如何通过可改变的风险因素（如饮食和运动）延迟是至关重要的。 来自德克萨斯理工大学的NSF资助的数学家和营养科学家正在共同开发新方法，用于研究肥胖和糖尿病中可改变的风险因素如何影响AD，例如人们选择吃的食物类型或运动量和睡眠量。 为此，研究人员正在开发新的数学模型，将大脑炎症和代谢应激等因素与AD进展联系起来，并生产必要的计算机软件，以解决由医学数据生成的大型复杂人脑图上的这些问题。 该研究旨在确定哪些可改变的因素对延迟AD最重要，研究人员团队正在与农村社区领导人合作，将他们的结果直接带入美国家庭的生活和餐桌上。该项目将为数学研究生和一名暑期兼职本科生提供培训。此外，该项目将允许研究人员与得克萨斯州农业生命推广服务合作，通过“德克萨斯人更好的生活”，将这项研究的结果直接带到拉伯克和周边县的服务不足人群中。氧化应激、脑胰岛素敏感性和神经炎症是阿尔茨海默病（AD）病理学的突出机制，也是已知的可改变的AD风险因素。 哪些可变因素介导了AD的发病和进展？在人类受试者中调查AD可改变风险因素的详细研究面临重大的伦理，技术或财务障碍。 为了避免这些挑战，研究小组将采用一种基于数学模型的敏捷方法，该模型描述了AD相关淀粉样蛋白β和tau蛋白病理学在由医学数据生成的复杂人脑网络上的演变。跨学科项目的目标有三个方面。 首先，设计新的，高维网络动力学系统（NDS），表达AD蛋白病理学的演变中存在的扰动胰岛素稳态，氧化应激和神经炎症。 第二，构建有效的计算软件，基于高性能库，如PETSc和SUNDIALS，通过可访问的高级编程接口实例化和解决大型NDS模型。 第三，使用微分方程理论，网络和数据科学的技术分析NDS模型，以及对来自患者神经成像数据的脑图进行AD病理学的大型计算模拟。 结合这些研究成果，将有助于了解改变可改变的AD风险因素的有效性，并为公众提供高水平的建议，以减轻他们一生中发展AD的风险。该奖项反映了NSF的法定使命，并被认为值得通过使用基金会的知识价值和更广泛的影响审查标准进行评估来支持。

Americans are losing the battle for minds. Obesity and type 2 diabetes accelerate the progression of deadly, age-related neurodegeneration like Alzheimer's disease (AD). In fact, by 2050, AD will impact every American family; 1 in 7 Americans over 65 will have AD, and there is no clear, effective pharmaceutical treatment regime in sight. In the absence of practical pharmaceutical treatment strategies for AD, understanding how AD can be delayed through modifiable risk factors, such as diet and exercise, is of paramount importance. NSF-funded mathematicians and nutritional scientists from Texas Tech University are coming together to develop new methods for studying how the modifiable risk factors in obesity and diabetes, like the types of food one chooses to eat or how much exercise and sleep one can get, affect AD. To do this, the investigators are developing new mathematical models to relate factors like brain inflammation and metabolic stress to AD progression and producing the necessary computer software to solve these types of problems on large, complex human brain graphs generated from medical data. The research aims to identify what modifiable factors matter the most for delaying AD, and the team of investigators is partnering with rural community leaders to bring their results straight into the lives, and onto the dinner tables, of American families. This project will provide training for graduate mathematics students and one part-time summer undergraduate. Additionally, this project will allow the investigators to partner with Texas A&M AgriLife Extension Service to bring the results of this research directly to under-served populations in Lubbock and surrounding counties through Better Living for Texans.Oxidative stress, brain insulin sensitivity and neuroinflammation are salient mechanisms of Alzheimer's disease (AD) pathology and known associates of modifiable AD risk factors. What mediation of modifiable factors delay the pathogenesis and progression of AD? Detailed studies that investigate modifiable risk factors for AD in human subjects face significant ethical, technical or financial barriers. To avoid these challenges, the research team will employ an agile approach based on mathematical models that describe the evolution of AD-associated amyloid-beta and tau protein pathology on complex human brain networks generated from medical data. The goals of the interdisciplinary project are threefold. First, to devise novel, high-dimensional network dynamical systems (NDS) that express the evolution of AD protein pathology in the presence of perturbed insulin homeostasis, oxidative stress and neuroinflammation. Second, to construct effective computational software, based on high-performance libraries such as PETSc and SUNDIALS, that instantiates and solves large NDS models via an accessible, high-level programming interface. Third, to analyze the NDS models using techniques from the theory of differential equations, networks and data science in addition to performing large computational simulations of AD pathology on brain graphs derived from patient neuroimaging data. Combining these research outcomes will develop an understanding of the efficacy of altering modifiable AD risk factors and enable high-level recommendations for the public to mitigate the risk of developing AD in their lifetime.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.