SBIR Phase I: Architectural Epidemiology: Leveraging Machine Learning and Spatial Data at Scale to Understand Health Outcomes

小型企业创新研究(SBIR)第一阶段项目的更广泛影响/商业潜力是能够利用架构作为促进健康和福祉的新工具。越来越多的证据表明，医院架构系统性地影响患者的健康、运营效率和患者满意度。然而，建筑师往往缺乏可靠的工具来预测设计选择将如何影响建筑运营。医院和诊所的设计往往与建筑物所有者无法实现组织指标和目标的情况不符，患者可能会看到更差的护理质量、更多的医疗差错，甚至更高的死亡率。商业影响将来自一个软件平台，该平台为架构师提供部署数据驱动设计所需的无代码工作流，同时管理项目时间表和内部成本。虽然第一阶段项目侧重于医疗体系结构，但更广泛的影响将是建筑师将能够设计所有建筑以优化健康、生产力和组织目标。这一小型企业创新研究(SBIR)第一阶段项目将支持以下算法的开发：1)生成结构化建筑数据以捕获影响居住者健康的空间品质的算法；2)用于联合探索和验证空间数据和健康数据的数据可视化接口；以及3)统计和机器学习模型，用于在大规模数据集中确定建筑特征和健康结果之间的有意义的关系。这些工具将通过对大型卫生系统进行概念验证研究来进行测试和评估。这项研究的目标是评估未经测试的统计机器学习方法的有效性，以确定体系结构和健康结果之间的有意义的关系。拟议的研究将为架构在影响健康结果方面的作用提供新的洞察力，为数据驱动的设计提供一种新的方法。虽然这项建议的重点是建筑数据，但探索的原则将通过1)将定性和定量数据纳入机器学习模型，2)通过数据可视化评估新的可解释性模式，以及3)改进处理稀疏但丰富的数据集的方法。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase I project is the ability to leverage architecture as a new tool for advancing health and wellbeing. A growing body of evidence demonstrates that hospital architecture systemically affects patient health, operational efficiency, and patient satisfaction. However, architects often lack the tools to reliably predict how design choices will impact building operations. Often hospitals and clinics designs fall short with building owners not able to achieve organizational metrics and goals, and patients can see poorer quality of care, increased medical errors, and even higher mortality rates. The commercial impact will stem from a software platform that equips architects with no-code workflows necessary to deploy data-driven design while managing project timelines and internal costs. Although the Phase I project focuses on healthcare architecture, the broader impact will be that architects will be able to design all buildings to optimize health, productivity, and organizational goals.This Small Business Innovation Research (SBIR) Phase I project will enable development of 1) algorithms that generate structured architectural data capturing spatial qualities that affect occupant health, 2) data visualization interfaces for exploring and validating spatial data and health data jointly, and 3) statistical and machine learning models for identifying meaningful relationships between architectural characteristics and health outcomes across large-scale datasets. These tools will be tested and evaluated by conducting a proof of concept study with a large health system. The goal of the study is to evaluate the efficacy of untested statistical machine learning methods to identify meaningful relationships between architecture and health outcomes. The proposed research will provide new types of insight into architecture's role in affecting health outcomes, providing a novel approach to data-driven design. Though the focus of this proposal is on architectural data, the principles explored will push boundaries of current limits by 1) incorporating qualitative and quantitative data in machine learning models, 2) assessing new modes of interpretability via data visualization, and 3) advancing methods for working with sparse but rich datasets.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.