SBIR Phase I: Developing Artificial intelligence Models to Predict In-hospital Clinical Trajectories for Heart Failure Patients

SBIR 第一阶段：开发人工智能模型来预测心力衰竭患者的院内临床轨迹

• 批准号: 2304358
• 负责人: Ruizhi Liao
• 金额: $ 27.5万
• 依托单位: EMPALLO, INC.
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-09-01 至 2024-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2304358&HistoricalAwards=false
• 关键词: SBIR; Phase; Developing; Artificial; intelligence

The broader/commercial impact of this Small Business Innovation Research (SBIR) Phase I project includes improving cardiovascular management, personalized medicine, inclusivity for historically underserved populations, and clinical trial design. The project could improve the health and wellbeing of heart failure (HF) patients while saving billions of dollars in HF hospitalization costs. If the technology proves feasible, it could shift the paradigm of HF management from reactive to proactive. The proposed machine learning model extracts latent features and detects subtle patterns from clinical data, which derives digital biomarkers that can potentially enable novel phenotype discovery and eventually personalized medicine. The digital biomarkers derived from the proposed innovation, when used in clinical trials, could also improve inclusivity and greater generalizability of novel therapies when applied to diverse populations. The proposed technology could enable clinical trial sponsors to achieve the desired statistical power with smaller patient populations. This, in turn, would enable faster, cheaper, and more effective clinical trials.This Small Business Innovation Research (SBIR) Phase I project mitigates the burden of heart failure (HF), which afflicts over 6.5 million Americans. As the leading cause of hospitalization in the U.S., HF results in more than $29 billion in hospital charges and $11 billion in hospitalization costs, annually. A large portion of hospitalization costs are driven by readmissions, with about 20% of heart failure patients readmitted within 30 days of discharge. The fundamental challenge is the variability of this disease. A treatment regimen that works for one patient might not work for another, even if they show similar symptoms. Anticipating clinical trajectories, treatment response, and potential complications, and translating those insights into actionable interventions is key to improving outcomes for HF patients. To help clinicians anticipate a HF patient’s response to treatment and adverse events during hospitalization and enable personalized intervention planning, this project will develop explainable and generalizable multimodal artificial intelligence (AI) models that predict a HF patient’s clinical trajectory shortly after admission. This technology is a methodological innovation grounded in large-scale, multi-center, clinical data. The key milestone in Phase I is to yield a reasonably accurate predictive AI model, cross-validated between the data of two large healthcare systems.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.

这项小企业创新研究（SBIR）第一阶段项目的更广泛/商业影响包括改善心血管管理、个性化医疗、对历史上服务不足人群的包容性以及临床试验设计。该项目可以改善心力衰竭（HF）患者的健康和福祉，同时节省数十亿美元的心力衰竭住院费用。如果这项技术被证明是可行的，它可能会将高频管理的模式从被动转变为主动。提出的机器学习模型从临床数据中提取潜在特征并检测细微模式，从而派生出可能实现新表型发现并最终实现个性化医疗的数字生物标志物。从提出的创新中获得的数字生物标志物，在临床试验中使用时，也可以在应用于不同人群时提高新疗法的包容性和更大的普遍性。所提出的技术可以使临床试验发起人在较小的患者群体中获得所需的统计能力。反过来，这将使临床试验更快、更便宜、更有效。这个小企业创新研究（SBIR）第一阶段项目减轻了心力衰竭（HF）的负担，这困扰着650多万美国人。作为美国住院治疗的主要原因，心衰每年导致290亿美元以上的住院费用和110亿美元的住院费用。住院费用的很大一部分是由再入院造成的，约20%的心力衰竭患者在出院后30天内再次入院。最根本的挑战是这种疾病的可变性。对一个病人有效的治疗方案可能对另一个病人无效，即使他们表现出相似的症状。预测临床轨迹、治疗反应和潜在并发症，并将这些见解转化为可操作的干预措施，是改善心衰患者预后的关键。为了帮助临床医生预测心衰患者在住院期间对治疗的反应和不良事件，并实现个性化的干预计划，该项目将开发可解释和可推广的多模式人工智能（AI）模型，用于预测心衰患者入院后不久的临床轨迹。这项技术是基于大规模、多中心、临床数据的方法论创新。第一阶段的关键里程碑是产生一个相当准确的预测人工智能模型，在两个大型医疗保健系统的数据之间进行交叉验证。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。