CAREER: Foundational statistical theory and methods for analyzing populations of attributed connectomes

由于大脑惊人的复杂性，了解它被称为科学的最后前沿。它为人们提供了无与伦比的认知能力，也是我们一些最虚弱的残疾的根源。脑科学家普遍认为，大脑本质上是一个复杂的大网络，网络的每个节点和边缘本身就是一个复杂的对象。因此，为了了解我们自己，并与包括抑郁症（世界上主要的疾病负担）和自杀（目前的流行病）在内的疾病作斗争，该项目的科学家将开发数学和统计工具来研究网络并将其应用于大脑网络。这将提供对人类大脑更深入的了解，并有助于预防功能障碍和恢复功能。他们开发的所有工具也将集成到课程中，所有的科学，包括论文、代码和教育材料，都将在开源中生成，这样社会上的每个人都可以访问这些内容。研究者将建立分析归属连接体种群的基础理论和方法。他们的方法，“连接编码”，将使大脑科学家能够(1)从单个连接组推断潜在结构，(2)在连接组群体中识别有意义的集群，(3)检测连接组和多变量表型之间的关系。他们开发的方法自然会克服连接组学固有的挑战：具有多层次非线性相互作用的高维非欧几里得数据。他们的程序将在理论保证、计算可扩展性、经验性能和结果可解释性方面扩展当前最先进的技术。与获得数据的领域专家合作，将在跨越实验模式、尺度和分类群的数据集上证明这些方法的有效性。他们还将产生教育资源，以补充他们的方法、数据和代码，使连接组学民主化。所有项目的成果将在https://neurodata.io/graspyThis上公布。奖项反映了NSF的法定使命，并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，被认为值得支持。

Understanding the brain has been called science’s final frontier due to its astonishing complexity. It provides people with their unmatched cognitive capabilities and is the source of some of our most debilitating disabilities. A widespread view among brain scientists is that brains are essentially large complex networks, with each node and edge of the network itself being a complex object. Thus, to understand ourselves and to fight disorders including depression (the world’s leading burden of disease) and suicide (a current epidemic), the scientist on this project will develop mathematical and statistical tools to study networks and apply them to brain networks. This will provide a deeper understanding of human brains and contribute to preventing dysfunction and restoring function. All the tools they develop will also be integrated into a course, and all of the science, including papers, code, and educational materials, will be generated in the open source, so that everyone in society will have access to this content. The investigator will establish foundational theory and methods for analyzing populations of attributed connectomes. Their approach, “connectal coding,” will enable brain scientists to (1) infer latent structure from individual connectomes, (2) identify meaningful clusters among populations of connectomes, and (3) detect relationships between connectomes and multivariate phenotypes. The methods they develop will naturally overcome the challenges inherent in connectomics: high-dimensional non-Euclidean data with multi-level nonlinear interactions. Their procedures will extend the current state-of-the-art in terms of theoretical guarantees, computational scalability, empirical performance, and interpretability of results. The efficacy of the methods will be demonstrated on datasets spanning experimental modalities, scales, and taxa, in collaboration with domain experts who acquired the data. They will also generate educational resources to complement their methods, data, and code to democratize connectomics. All project results will be available at https://neurodata.io/graspyThis 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.