Understanding and Predicting High-Need, High-Cost Patients among Older Adults

项目摘要/摘要 这是由国家老龄研究所提交的K99/R00独立之路奖 张永康，威尔康奈尔大学医疗政策和研究部副研究员 医学院(WCMC)。张博士的职业目标是成为改善护理的独立研究员 为老年人开发和应用有效的预测工具来识别老年人的复杂性 条件和高度的医疗保健需求。此K99/R00应用程序将为张博士提供必要的 培训1)了解高需求、高成本(HNHC)老年人的复杂性和特点；2) 使用机器学习方法开发HNHC成人预测模型；3)测试性能 基于机器学习的模型和三种常用的患者风险预测工具。 张博士组建了一支由多个部门的资深研究人员组成的导师团队，并 威尔·康奈尔医学院系：纳内特·莱特曼的Rainu Kaushal博士(主要导师) HNHC患者和健康数据科学方面的杰出教授和专家；Lawrence Casalino博士 (共同导师)他是利文斯顿·法兰德公共卫生教授，也是 为HNHC患者提供医疗服务；马克·拉克斯博士(共同导师)，老年病学和外科教授- Ticing老年病学家，老年人复杂性和医疗保健需求方面的专家；包玉华博士(联合 导师)，医疗保健政策和研究副教授，行为健康专家 条件和处方数据；王飞博士(共同导师)，健康信息副教授- ICS和机器学习方法专家；以及助理教授James Flory博士(顾问)- 医疗保健政策和研究及执业临床医生和用药专家。 HNHC老年人是一小部分患者，占医疗保健用途的比例不成比例- 规模化。这些患者更有可能遇到可预防的质量和安全问题，因为他们的 与卫生系统的即时互动。照顾HNHC老年人为QUAL提供了巨大的潜在好处- 城市提升和成本降低。然而，除非这些患者能够做到，否则这些好处不太可能实现 被正确识别和定位。基于他之前关于开发索赔的研究和培训 对于HNHC医疗保险患者的基于数据的分类，张博士的研究将理解 HNHC老年人，并使用临床和处方数据开发这些患者的预测因素(目标1)， 建立了一个基于机器学习的HNHC老年人预测模型(目标2)，并比较了其性能。 使用三种常用的患者风险预测工具对预测模型进行分析(目标3)。这项研究 将是R01拨款申请的基础，该申请将把此预测模型纳入医疗保健去中心化- 统一流程，并确定最佳方式，以告知改善医疗保健服务的机会。

PROJECT SUMMARY/ABSTRACT This is a K99/R00 Pathway to Independence Award submitted to the National Institute on Aging by Yongkang Zhang, a Research Associate in the Department of Healthcare Policy and Research at Weill Cornell Medical College (WCMC). Dr. Zhang’s career goal is to become an independent researcher on improving care for the elderly through developing and applying effective prediction tools to identify older adults with complex conditions and high healthcare needs. This K99/R00 application will provide Dr. Zhang with the necessary training 1) to understand the complexity and characteristics of high-need, high-cost (HNHC) older adults; 2) to develop a prediction model for HNHC adults using machine learning methods; and 3) to test the performance of the machine learning-based model with three commonly used, patient-risk prediction tools. Dr. Zhang has assembled a mentor team of accomplished researchers across multiple divisions and departments at Weill Cornell Medical College: Dr. Rainu Kaushal (primary mentor) who is the Nanette Laitman Distinguished Professor and an expert on the HNHC patients and health data science; Dr. Lawrence Casalino (co-mentor) who is the Livingston Farrand Professor of Public Health and an expert on characteristics of and healthcare delivery for HNHC patients; Dr. Mark Lachs (co-mentor) who is a Professor of Geriatrics and prac- ticing geriatrician and an expert on the complexity and healthcare needs of older adults; Dr. Yuhua Bao (co- mentor) who is an Associate Professor of Healthcare Policy and Research and expert on behavioral health conditions and prescription data; Dr. Fei Wang (co-mentor) who is an Associate Professor of Health Informat- ics and an expert on machine learning methods; and Dr. James Flory (consultant) who is an Assistant Profes- sor of Healthcare Policy and Research and practicing clinician and an expert on medication use. HNHC older adults are small group of patients representing a disproportionate share of healthcare utili- zation. These patients are more likely to experience preventable quality and safety problems due to their fre- quent interactions with health systems. Caring for HNHC older adults provides great potential benefits for qual- ity improvement and cost reduction. However, the benefits are unlikely to be realized unless these patients can be correctly identified and targeted. Building on his previous research and training on developing a claims data-based taxonomy for HNHC Medicare patients, Dr. Zhang’s research will understand the characteristics of HNHC older adults and develop predictors for these patients using clinical and prescription data (Aim 1), de- velop a machine learning-based prediction model for HNHC older adults (Aim 2), and compare the perfor- mance of the prediction model with three commonly used, patient risk prediction tools (Aim 3). This research will be the foundation for an R01 grant application that will incorporate this prediction model into healthcare de- livery process and identify the optimal ways to inform opportunities for improvement in healthcare delivery.