Statistical Methodology and Applications to Engineering and Economics

在2010年医疗保健和金融改革立法之后，过去三年见证了金融市场和美国医疗保健系统新时代的开始。研究的长期目标是开发创新的统计方法，并将其与高性能计算和通信网络的进步联合收割机相结合，以应对新时代定量金融和医疗保健的挑战。该研究将导致生物医学，经济学和工程统计方法的进步和创新，为及时应用于临床和转化医学研究，医疗保健，国土安全，环境变化和风险管理铺平道路。通过让研究生参与研究的各个阶段，以及围绕研究及其应用开发新的课程材料，培养学术界、产业界和政府的下一代科学家，这一研究产生了更广泛的影响。研究项目大致可分为四个领域。第一个是具有协变量的多臂强盗，在机器学习中也被称为“上下文强盗”，以及它们在医学和电子商务中的个性化策略中的应用，特别是在神经科学中基因组指导的个性化癌症治疗和生物标志物指导的抑郁症治疗中的应用。第二个领域是制造系统网络模型中的故障检测和监控，用于金融市场的系统性风险，国土安全以及环境或全球变化。第三个领域是有效的自适应粒子滤波器在非线性状态空间模型，在工程和经济学中有广泛的应用，与机器人和随机自适应控制作为本研究的主要焦点的发展。第四个领域包括信贷市场中多家公司联合违约风险的动态经验贝叶斯建模，零售银行业务和保险索赔的贷款损失风险，宏观经济时间序列建模和预测，以及医疗保健成本和预防管理的数据分析。

The past three years witnessed the beginning of a new era in financial markets and in the US health care system, following the health care and financial reform legislation in 2010. A long-term objective of the research is to develop innovative statistical methodologies and to combine them with advances in high-performance computing and communication networks for addressing the challenges in quantitative finance and health care in this new era. The research will lead to advances and innovations in statistical methods in biomedicine, economics and engineering, paving the way for timely applications to clinical and translational medical research, health care, homeland security, environmental change, and risk management. A broader impact of this research is the training of the next generation of scientists in academia, industry, and government, by involving graduate students in all phases of the research, and by developing new course material built around the research and its applications.The research projects can be broadly divided into four areas. The first is multi-arm bandits with covariates, also called "contextual bandits" in machine learning, and their applications to personalized strategies in medicine and electronic business, in particular, to genomic-guided personalized cancer treatments and biomarker- guided treatments for depression in neuroscience. The second area is fault detection and surveillance in network models for manufacturing systems, for systemic risk in financial markets, for homeland security, and for environment or global change. The third area is the development of efficient adaptive particle filters in nonlinear state-space models that have far-ranging applications in engineering and economics, with robotics and stochastic adaptive control as the main focus in this research. The fourth area includes dynamic empirical Bayes modeling of joint default risk for multiple firms in credit markets, of loan loss risk in retail banking and of insurance claims, macroeconomic time series modeling and forecasting, and data analytics for health care cost and preventive management.