Technology Application to Enhance Discharge Referral Decision Support

描述（由申请人提供）：通过更好的出院计划（DP）和过渡性护理减少再入院是国家医疗保健的优先事项。RightCare Solutions利用了由护士研究人员领导的10多年跨学科学术研究，并通过我们非常成功的第一阶段SBIR赠款证明了D2 S2产品的市场价值和我们团队的技术专长。D2 S2是RightCare Solutions在医院EHR中安装的六项决策支持工具，用于协助出院计划人员在入院时识别高风险患者，以便有时间和精力针对适当的过渡期和急性期后护理，以防止再次入院。我们已经从第一阶段的奖项中取得了出色的成果，并正在提出进一步的技术发展，以加强我们的商业推出。D2 S2工具的市场潜力很大，因为出院决策支持估计适用于大约6，500家美国医院，其中60%为老年人，相当于每年1400万次出院。第1阶段的结果表明，对30天和60天的再入院有显著影响，这为我们提供了产品价值的有力证据。然而，结果和我们使用该软件的经验表明，有机会提高产品的应计性和功能。我们建议通过创新的数据挖掘和机器学习技术来提高我们的预测准确性，并通过电子连接急性和急性后护理环境来改善功能。由于在宾夕法尼亚大学卫生系统的三家医院实施，我们有超过6,000名患者的数据，通过持续的实时实施，我们将在第2阶段拨款开始时积累超过25,000名患者的数据。利用现有的和新的数据产生的持续运作，这一拟议的SBIR赠款将推动该产品在两个主要方面，以提高其用户的商业利益。目标1：开发、测试和扩展SMART功能，这是一个通过使用医院特定和患者水平特征（D2 S2变量和其他临床和非临床特征）以及现代数据挖掘/机器学习技术来提高预测准确性的动态过程。目标2：通过电子方式将高风险患者与急性期后护理（PAC）提供者和利益相关者联系起来，将D2 S2建议付诸实施，这被称为ACCECT功能。被称为“学习健康系统”的最终产品，这SBIR赠款将提供持续的评估和改进，以最终用户衡量他们的目标。我们的创新设计产生了一个闭环系统，该系统将随着时间的推移使用来自D2 S2和医院数据库的数据，以“变得更智能”。“此外，我们的增强型产品将把急性护理与急性后提供者联系起来，向他们提前警告即将前来接受护理的患者。

DESCRIPTION (provided by applicant): Decreasing readmissions through better discharge planning (DP) and transitional care is a national healthcare priority. RightCare Solutions has leveraged over 10 years of interdisciplinary academic research led by a nurse researcher, and through our highly successful phase one SBIR grant demonstrated market value for the D2S2 product and the technical expertise of our team. The D2S2 is a six-item decision support tool installed by RightCare Solutions in the hospital EHR to assist discharge planners to identify high-risk patients upon admission allowing time and focus to target appropriate transitional and post-acute care to prevent readmissions. We have achieved outstanding outcomes from our phase 1 award and are proposing further technological developments to enhance our commercial launch. The market potential for the D2S2 tool is significant since discharge decision support is estimated to be applicable to roughly 6,500 U.S. hospitals with a census that is 60% older adults equaling 14 million discharges per year. The phase 1 results indicate a significant impact on 30 and 60 day readmissions giving us strong evidence as to the value of the product. However, the results and our experience using the software indicate there is opportunity to enhance the product's accruacy and functionality. We propose to enhance our predictive accuracy through innovative data mining and machine learning techniques and to improve the functionality by electronically connecting the acute and post-acute care settings. Due to implementation in the three hospitals of the University of Pennsylvania Health System we have data on over 6,000 patients and through the continued live implementation we will accumulate data on over 25,000 patients by the start of the phase 2 grant. Using existing, and new data generated from continued operations, this proposed SBIR grant will advance the product in two major ways to enhance the commercial benefit to its users. Aim 1: Develop, test, and scale SMART capabilities, a dynamic process for improving prediction accuracy, by using hospital-specific and patient level characteristics (D2S2 variables and additional clinical and non-clinical characteristics) and modern data mining/machine learning techniques. AIM 2: Operationalize the D2S2 recommendations by electronically connecting high-risk patients with post-acute care (PAC) providers and stakeholders, known as CONNECT capabilities. Called a "learning health system" the end- product of this SBIR grant will provide continuous evaluation and improvement to the end-user measured against their goals. Our innovative design produces a closed-loop system that will use data from the D2S2 and hospital databases over time to "get smarter." Further, our enhanced product will link acute care to post-acute providers giving them advanced warning of patients who will shortly come to them for care.