I-Corps: Machine Learning Algorithm for Cardiovascular Disease Diagnosis

I-Corps项目更广泛的影响/商业潜力是开发心血管筛查/诊断软件工具，使临床医生能够为所有患者群体提供更高质量的护理，包括历史上服务不足的患者和心血管疾病风险范围内的患者。女性心血管疾病的误诊率比男性高出50%。该软件旨在提高临床医生的诊断准确性，并为患者提供更好的心血管健康结果。心血管疾病是全球死亡的主要原因，该软件有可能通过减少不必要的测试，程序和疾病的支出来降低心血管医疗保健成本，因为患者没有及早筛查和快速准确地诊断。这项技术的重点是那些经常被误诊的心血管疾病，鉴于妇女被误诊的风险增加，以及这些疾病对影响少数性别和种族患者的心血管健康不平等的持续负担的贡献，旨在改善卫生公平。I-Corps项目的基础是开发一种基于机器学习的算法，以帮助临床医生识别哪些患者有发展或恶化心血管疾病的高风险。拟议的技术使用患者电子健康记录中的信息，算法侧重于不太了解、经常被遗漏和/或对女性影响过大的心血管疾病。机器学习（ML）算法在高度多样化的患者群体的数字健康记录上进行了训练和测试，与目前的护理标准相比，它可能更准确地为女性和少数民族提供心血管疾病（CVD）诊断。大多数用于诊断心血管疾病的机器学习工具都使用深度学习来自动解释图像，并以与专科医生相似或更高的精度解释心电图（ECG）信号。用于该技术的培训模型旨在发现遗漏的心血管疾病病例，并基于患者电子健康记录中常见的信息。这是因为在许多误诊病例中，临床医生没有怀疑CVD，没有安排CVD特异性检查/扫描或信号显示正常。该技术利用了患者的生理差异，旨在提高患者分诊的准确性，并能够比现有的基于规则的系统更早地识别出心血管疾病患者。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。

The broader impact/commercial potential of this I-Corps project is the development of a cardiovascular screening/diagnostic software tool that empowers clinicians to deliver higher quality care to all patient populations, including the historically underserved and those across the spectrum of cardiovascular disease risk. Cardiovascular disease misdiagnosis is as much as 50% higher in women than men The proposed software is designed to improve diagnostic accuracy for clinicians and provide better cardiovascular health outcomes for patients. Cardiovascular disease is the leading cause of death globally, and this software has the potential to reduce cardiovascular healthcare costs by reducing spending on unnecessary testing, procedures, and illness that occurs when patients are not screened early and diagnosed quickly and accurately. This technology focuses on those cardiovascular conditions that are misdiagnosed frequently and aims to improve health equity given the increased risk of misdiagnosis for women and the contribution of these conditions to the ongoing burden of cardiovascular health inequities that affect patients from minoritized genders and ethnicities.This I-Corps project is based on the development of a machine learning-based algorithm to assist clinicians in identifying which patients are at high risk of developing or worsening cardiovascular diseases. The proposed technology uses information in patients’ electronic health records, and the algorithm focuses on cardiovascular diseases that are not well understood, are often missed, and/or disproportionately affect women. The machine learning (ML) algorithm was trained and tested on the digital health records of a highly diverse group of patients and may more accurately provide cardiovascular disease (CVD) diagnoses for women and ethnic minorities than the current standard of care. Most ML tools for diagnosing CVD use deep learning to automate the interpretation of images and to interpret electrocardiogram (ECG) signals with similar or superior accuracy to specialist physicians. The training model used for this technology is designed to catch missed cases of CVD and is based on information that is commonly in patients’ electronic health records. This is because in many cases of misdiagnosis, CVD is not suspected by the clinician and CVD-specific tests/scans are not ordered or signals appear normal. The technology leverages physiological differences in patients and is developed with the aim of improving the accuracy of triaging patients and being able to identify patients with CVD earlier than with the existing rule-based 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.