SBIR Phase I: Advanced computational methods for forecasting multiple types of economic and social returns

这个小企业创新研究（SBIR）第一阶段项目的更广泛的影响/商业潜力是为社会企业提供使用机器学习、计算机模拟和数据可视化执行社会影响预测和跟踪的软件。这些数字工具为社会企业提供了更好的可视性，使他们能够更好地了解他们的资源可以投资在哪里以产生最大的影响。定量数据分析的采用使商业企业变得更加强大，但这些工具在非营利组织、慈善基金会、国际发展机构、影响力投资者和政府机构的7万亿美元世界中进展缓慢。该软件通过提供每美元影响的定量估计来通知投资策略。该软件还将跟踪每项资源分配的持续社会影响，使社会企业能够不断适应和提高其运作效率，并对其资金来源提供更多的问责制。这些软件工具可以帮助解决一些最令人烦恼的社会问题，而不是跨越健康、可持续性、贫困和教育机会。这个小企业创新研究（SBIR）第一阶段项目利用软件数据分析工具，包括基于代理的建模（ABM）、机器学习（如梯度增强机器和人工神经网络）和交互式数据可视化。基于Agent的建模包括建立目标生态系统的数字模拟，捕捉基本的参与者和行为。例如，对于阿片类药物流行病，这可能是成瘾者/药物/医生/制药代理商，以及他们的属性、历史和相互作用。该模型允许模拟干预方案和量化结果。机器学习技术一旦经过目标生态系统的过去数据训练和校准，就可以提供预测并允许探索假设场景。针对每个目标社会问题建立一个自下而上的模型，由相关社会企业共享，并利用他们的集体数据和主题专家进行校准。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 Small Business Innovation Research (SBIR) Phase I project is providing social enterprises with software that performs social-impact forecasting and tracking using machine learning, computer simulation, and data visualization. These digital tools provide social enterprises with better visibility into where their resources can be invested for maximum impact. The adoption of quantitative data analysis has made commercial enterprises more formidable, but these tools are slow in coming to the $7 trillion world of non-profits, charitable foundations, international development agencies, impact investors, and government agencies. The software informs investment strategies by providing a quantitative estimate of impact-per-dollar. The software will also track the ongoing social impact of each resource allocation, enabling social enterprises to iteratively adapt and improve the efficacy of their operations, and provide more accountability to their funding sources. These software tools can help address some of the most vexing societal concerns than span health, sustainability, poverty, and access to education.This Small Business Innovation Research (SBIR) Phase I project makes use of software data analysis tools including Agent Based Modeling (ABM), machine learning (such as Gradient Boosted Machines and Artificial Neural Networks), and Interactive Data Visualization. Agent Based Modeling involves building a digital simulation of the target ecosystem, capturing the essential actors and behaviors. For example, for the opioid epidemic, this may be the addicts/drugs/doctors/pharma agents, along with their attributes, histories, and interactions with each other. This modeling allows simulation of intervention scenarios and quantification of outcomes. Machine learning techniques, once trained and calibrated with past data on the target ecosystem, provide forecasting and allow the exploration of what-if scenarios. A bottom-up model is built for each target social issue, shared by relevant social enterprises, and calibrated using their collective data and subject matter experts.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.