Automated point-of-care identification of innocent Still's murmur in children

自动即时识别儿童无辜斯蒂尔杂音

• 批准号: 9903425
• 负责人: Robin Doroshow
• 金额: $ 70.89万
• 依托单位: AUSCULTECH DX, LLC
• 依托单位国家: 美国
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-09-06 至 2024-03-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9903425
• 关键词: Algorithms; Anxiety; Apple; Benign; Bluetooth; Boston; Cellular Phone; Characteristics; Child; Childhood; Classification; Clinical; Collaborations; Computer software; Congenital Heart Defects; Consultations; Data; Decision Support Systems; Devices; Diagnosis; Emotional; Evaluation; Exhibits; Family; Future; Goals; Healthcare; Healthcare Systems; Hearing; Heart Diseases; Heart Sounds; Heart murmur; Incidence; Legal patent; Libraries; Medical center; Military Personnel; Modeling; Multicenter Trials; Patients; Pattern; Pediatric Hospitals; Performance; Phase; Physicians; Population; Research; Resources; Sample Size; Small Business Technology Transfer Research; Specialist; Specificity; Stethoscopes; Technology; Technology Transfer; Testing; Training; United States; Validation; Wireless Technology; Work; base; care providers; classification algorithm; clinical center; cloud based; commercialization; computerized; cost; design; digital; handheld mobile device; improved; machine learning algorithm; mobile computing; new technology; novel; pediatric cardiologist; performance tests; point of care; prototype; smartphone Application; sound; tool; wasting; web interface

PROJECT SUMMARY The overall goal of the proposed research and technology transfer effort is to develop a low-cost, point-of-care, mobile device-based technology for automated identification of Still's murmur, the most common innocent (benign and harmless) heart murmur of childhood. This novel technology could reduce the current rate of over half a million children unnecessarily referred to pediatric cardiologists by primary care providers each year in the United States. In Phase I, AusculTech Dx and Children's National Medical Center developed a machine-learning algorithm capable of identifying Still's murmur with high accuracy (90% sensitivity, 99% specificity). To enable digital recording, essential for computerized murmur analysis, we also developed a digital stethoscope that connects to a smartphone and is equivalent in performance to a commercial stethoscope. Our envisioned product, called StethAid, is a combination of this novel digital stethoscope and a smartphone application (app). Having successfully met the Phase I performance milestones, we now propose to fully develop StethAid and conduct a multicenter trial through the following three specific aims: (1) develop and test a clinical-grade digital stethoscope, (2) develop fully feature software app, (3) conduct a multicenter trial of automated Still's murmur identification. For the multicenter trial, we have expanded our collaboration to include Boston Children's Hospital and Walter Reed National Military Medical Center. Our deliverable for Phase II is a technology platform validated by leading pediatric cardiologists that is ready for seeking regulatory approvals, deployment at PCP offices, and commercialization. As a decision support system, StethAid could empower PCPs to identify Still's murmur accurately and thus reduce the huge number of unnecessary specialist referrals. This should save the healthcare system hundreds of millions of dollars annually, allow pediatric cardiologists to focus on patients with serious conditions, and protect healthy children and their families from the unnecessary anxiety, inconvenience, and expense of seeing a pediatric cardiologist.

项目总结 拟议的研究和技术转让努力的总体目标是开发一种低成本、 基于移动设备的护理点技术，用于自动识别STEST的杂音， 童年时期最常见的天真(良性和无害)心脏杂音。这部小说 技术可以降低目前超过50万儿童不必要地被转诊的比率 在美国，儿科心脏病专家每年都会被初级保健提供者淘汰。在第一阶段， AusculTech Dx和国家儿童医学中心开发了一种机器学习算法 能够以较高的准确率(90%的灵敏度，99%的特异度)识别STER的杂音。至 实现数字录音，对于计算机化的杂音分析至关重要，我们还开发了一种 连接到智能手机的数字听诊器，在性能上相当于 商用听诊器。我们设想的产品，叫做Stethaid，是这部小说的组合 数字听诊器和智能手机应用程序(APP)。成功满足第一阶段的要求 绩效里程碑，我们现在建议全面开发Stethaid并进行多中心 通过以下三个具体目标进行试验：(1)开发和测试临床级数字 听诊器，(2)开发全功能软件APP，(3)进行自动化的多中心试验 斯蒂尔的杂音识别。对于多中心试验，我们已将合作范围扩大到 包括波士顿儿童医院和沃尔特里德国家军事医疗中心。我们的 可交付的第二阶段是一个由领先的儿科心脏病专家验证的技术平台， 已经准备好寻求监管部门的批准，在PCP办公室部署，并实现商业化。AS 作为一个决策支持系统，Stethaid可以使PCP能够准确地识别STER的杂音 从而减少大量不必要的专家转介。这应该会节省 医疗系统每年耗资数亿美元，使儿科心脏病专家能够专注于 ，并保护健康的儿童及其家人免受 不必要的焦虑、不便和看儿科心脏病专家的费用。

项目成果

Automated identification of innocent Still's murmur using a convolutional neural network.

• DOI: 10.3389/fped.2022.923956
• 发表时间: 2022
• 期刊: FRONTIERS IN PEDIATRICS
• 影响因子: 2.6
• 作者: Shekhar, Raj;Vanama, Ganesh;John, Titus;Issac, James;Arjoune, Youness;Doroshow, Robin W.
• 通讯作者: Doroshow, Robin W.