An automated AI/ML platform for multi-researcher collaborations for a NIH BACPAC funded Spine Phenome Project

项目摘要/摘要 慢性下腰痛(CLBP)是一种影响全球数百万人的衰弱疾病。尽管 由于干预措施使用率的增加和医疗费用的上升，慢性下腰痛的患病率继续上升。这 出现问题是因为慢性腰痛是复杂的、异质性的，而目前的诊断和治疗主要依赖于 关于主观指标，而不是针对与以下方面相关的所有多维生物心理社会机制 CLBP。具体地说，大多数诊断没有定量地考虑患者的功能测量。要解决这个问题 问题是，我们的家长基金专注于开发和验证数字健康平台，以提供 有意义的数据驱动的指标，支持对临床评估和治疗的集成方法 CLBP。然而，它并没有直接解决相关的运营和质量管理挑战 与多个研究伙伴一起大规模执行AI/ML分析。虽然最初的拨款将支持 一些人工AI/ML分析作为其交付内容的一部分，需要额外的工作才能实现AI/ML就绪 并最大限度地发挥BACPAC联盟和其他NIH开发的数据集的全部潜力 合作者。拟议的补充赠款将发展所需的基础设施和管道，以支持 以一种对AI/ML更友好的方式共享数据，同时还允许与其他 对AI/ML感兴趣的财团成员。技术开发工作将与合作伙伴合作完成 使用AWS云服务。具体目标是：1)开发AI/ML计算数据访问流水线 用于数字健康平台；以及2)开发将简化AI/ML工作流的平台功能。这个 该项目的结果将减少持续改革数据集所花费的管理费用时间，因此ML 研究人员可以专注于开发模型和确定可行的发现。总体而言，这一努力与 NIH在数据科学方面的战略目标，并有可能转变临床实践范式，改进 患者结果、提高护理效率和降低成本。

Project Summary/Abstract Chronic Low Back Pain (cLBP) is a debilitating condition that affects millions of people globally. Despite increased utilization of interventions and rising medical costs, cLBP prevalence has continued to increase. This problem arises because cLBP is complex, heterogeneous and current diagnostics and treatments rely primarily on subjective metrics and do not target all the multidimensional biopsychosocial mechanisms associated with cLBP. Specifically, most diagnostics do not quantitively consider patient functional measures. To address this problem, our parent grant is focused on developing and validating a digital health platform to provide meaningful data-driven metrics that enables an integrated approach to clinical evaluation and treatment of cLBP. However, it does not directly address the operational and quality management challenges associated with performing AI/ML analyses at scale with multiple research partners. While the original grant will support some manual AI/ML analyses as part of its deliverables, additional effort is needed to achieve AI/ML readiness and maximize the full potential of the developed dataset for the BACPAC consortium and other NIH collaborators. The proposed supplementary grant will develop the needed infrastructure and pipeline to support data sharing in a more AI/ML-friendly way, while also allowing for more intimate collaborations with other consortium members who are interested in AI/ML. Technology development effort will be done in partnership with AWS Cloud services. The specific aims are to: 1) develop an AI/ML Computational data access pipeline for the Digital Health Platform; and 2) develop platform features that will streamline AI/ML workflows. The outcome of this project will decrease the overhead time spent on constantly reforming datasets so ML researchers can focus on developing models and identifying actionable findings. Collectively, this effort aligns with NIH’s strategic goals for data science and has the potential to shift clinical practice paradigms, improve patient outcomes, enhance care efficiency, and reduce costs.