PFI-TT: Artificial Intelligence System for Enterprise Performance Management that Integrates Causal Analytics and Human Expertise

这个创新伙伴关系-技术转化（PFI-TT）项目的更广泛的影响/商业潜力是对企业绩效管理（EPM）中可解释的规定性（即因果关系）分析的未满足的市场需求作出反应。如果成功，提议的平台将使组织能够解决复杂的多目标管理问题，包括客户、员工和其他利益相关者的财务结果和福祉。建议的规定性分析解决方案有望通过改进企业范围的性能来优化决策。预计到2026年，EPM市场的销售额将达到77亿美元，但目前只有11%的大中型企业使用某种形式的规定性分析。值得注意的是，超过50%的医疗保健系统对目前可用的EPM软件不满意，这导致患者满意度和运营利润率下降。该项目更广泛的社会影响将帮助企业直观地监测经济绩效（如投资回报）与福祉结果（如客户和员工满意度）之间的权衡和协同效应，并确定最佳行动。本项目旨在帮助组织处理多利益相关者、多目标绩效管理的复杂性。它构建了医疗保健系统中的第一个商业用例，重点关注医疗保健组织的多利益相关者绩效管理。该技术将开发基于知识图的推理人工智能网络，这将使人机协作能够加速将领域知识从专家文档合成为因果知识图，并将这些图摄取到因果推理和数据科学工具中。所提议的项目的智力价值源于建立一个高性能的机器阅读系统，以从学术和行业文档中提取和综合因果见解。该团队还寻求为企业管理开发第一个基于知识图的推理系统，以识别隐藏的因果关系并提出干预建议。最后，该项目将开发一种新的方法来改进数据科学中的因果推理。通过将因果知识图与大量医疗保健系统数据样本相结合，该产品将根据每个医疗保健系统的观察站数据创建一个合成的反事实，并运行因果分析，以确定针对一组给定绩效目标的最有效行动。该奖项反映了美国国家科学基金会的法定使命，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。

The broader impact/commercial potential of this Partnerships for Innovation - Technology Translation (PFI-TT) project is to respond to the unmet market need for explainable prescriptive (i.e., causal) analytics in enterprise performance management (EPM). If successful, the proposed platform will enable organizations to solve complex, multi-objective management problems, including financial outcomes and well-being for customers, employees, and other stakeholders. The proposed prescriptive analytics solution is expected to optimize decision-making through improvements in enterprise-wide performance. The market for EPM is expected to reach $7.7 billion in sales by 2026, but only 11% of large and medium-sized enterprises currently use some form of prescriptive analytics. Notably, more than 50% of healthcare systems are unsatisfied with currently available EPM software, which leads to decreases in patient satisfaction and operating margins. This project's broader societal impact will help enterprises visually monitor the trade-offs and synergies between economic performance (e.g., return on investment) and well-being outcomes (e.g., customer and employee satisfaction) and identify optimal actions.This project seeks to help organizations deal with the complexity of multi-stakeholder, multi-objective performance management. It builds the first commercial use case in healthcare systems, focusing on healthcare organizations' multi-stakeholder performance management. The technology will develop a Knowledge Graph-Based Reasoning Artificial Intelligence Network, which will enable human-machine collaboration to accelerate the synthesis of domain knowledge from expert documents into causal knowledge graphs and ingest such graphs into causal inference and data science tools. The intellectual merit of the proposed project stems from building a high-performance machine reading system to extract and synthesize causal insights from both academic and industry documents. The team also seeks to develop the first knowledge graph-based reasoning system for enterprise management to identify hidden causal relations and make intervention recommendations. Finally, the project will develop a novel approach to improving causal inference in data science. By combining causal knowledge graphs with a large sample of healthcare system data, the product will create a synthetic counterfactual from observatory data for each healthcare system and run causal analysis to identify the most effective actions for a given set of performance objectives.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.