摘要
《机器人与机器学习每日新闻》的一名新闻记者兼新闻编辑——心脏病和疾病的新研究——他的艺术疾病是一篇报道的主题。根据NewsRx通讯员从荷兰Groni Ngen传来的消息，研究表明：“人工瓣膜心内膜炎(PVE)是人工瓣膜植入的严重并发症，估计年发病率至少为0.4%-1.0%。Duke Cr Iteria和随后的改进已被发展为感染性心内膜炎(IE)的诊断框架。”我们的新闻记者引用了罗宁根大学的研究，“然而，它们的敏感性和特异性有限，特别是FO R PVE。此外，它们的最新版本(ESC2015和ESC2023)包括先进的成像方式，如心脏CTA和[F]FDG PET/CT作为主要标准。然而，尽管有这些重大变化，使用主要和次要标准的Weiging系统保持不变。这可能给诊断标准集带来了偏差。在这里，我们旨在通过机器学习算法评估和改进修正的Duke/ESC 2015(MDE2015)Criter IA的预测值。在这项概念验证研究中，我们使用了一个定义明确的回顾性多中心队列的数据，对160例疑似PVE患者进行了评估，并将四种机器学习算法与MDE2015标准的诊断结果进行了比较：Lasso Logistic回归，梯度提升决策树(XGBoost)，梯度提升决策树，不梯度提升决策树，以及结合这些预测的模型(集合学习)。LL模型使用了同样的特征，也构成了MDE2015标准。PVE的最终诊断是基于心内膜炎团队的共识，使用所有可用的临床信息，包括无论何时进行手术发现，并至少随访1年，MDE2015标准的诊断性能因“可能”PVE病例的处理方式而异，将这些病例视为PVE阳性，敏感性和特异性分别为0.96和0.60，而将这些病例视为阴性，敏感性和特异性分别为0.74和0.98.结合ROCANTIC分析可能的心内膜炎为阳性和阴性的方法，Machine学习模型的AUC分别为0.917.，其敏感性和特异性分别为：Logisti C回归0.92和0.85；XGBoost，0.90和0.85；决策树，0.88和0.86；集成学习，0.91和0.85，得到的AUC分别为0.937，0.938.在这项概念验证研究中，机器学习算法与MDE2015 Criter IA中使用的主要/次要称重系统相比，取得了更好的诊断性能。此外，这些模型提供了可量化的诊断的确定性水平，潜在地提高了临床医生的可解释性。此外，它们都易于纳入新的和/或改进的标准，如CTA或[F]FDG PET/CT等先进成像方式的个体重量。
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News – New research on Heart Disorders and Diseases - He art Disease is the subject of a report. According to news originating from Groni ngen, Netherlands, by NewsRx correspondents, research stated, “Prosthetic valve endocarditis (PVE) is a serious complication of prosthetic valve implantation, w ith an estimated yearly incidence of at least 0.4-1.0%. The Duke cr iteria and subsequent modifications have been developed as a diagnostic framewor k for infective endocarditis (IE) in clinical studies.” Our news journalists obtained a quote from the research from the University of G roningen, “However, their sensitivity and specificity are limited, especially fo r PVE. Furthermore, their most recent versions (ESC2015 and ESC2023) include adv anced imaging modalities, e.g., cardiac CTA and [F] FDG PET/CT as major criteria. However, despite these significant changes, the we ighing system using major and minor criteria has remained unchanged. This may ha ve introduced bias to the diagnostic set of criteria. Here, we aimed to evaluate and improve the predictive value of the modified Duke/ESC 2015 (MDE2015) criter ia by using machine learning algorithms. In this proof-of-concept study, we used data of a welldefined retrospective multicentre cohort of 160 patients evaluat ed for suspected PVE. Four machine learning algorithms were compared to the pred iction of the diagnosis according to the MDE2015 criteria: Lasso logistic regres sion, decision tree with gradient boosting (XGBoost), decision tree without grad ient boosting, and a model combining predictions of these (ensemble learning). A ll models used the same features that also constitute the MDE2015 criteria. The final diagnosis of PVE, based on endocarditis team consensus using all available clinical information, including surgical findings whenever performed, and with at least 1 year follow up, was used as the composite gold standard. The diagnost ic performance of the MDE2015 criteria varied depending on how the category of ‘ possible’ PVE cases were handled. Considering these cases as positive for PVE, s ensitivity and specificity were 0.96 and 0.60, respectively. Whereas treating th ese cases as negative, sensitivity and specificity were 0.74 and 0.98, respectiv ely. Combining the approaches of considering possible endocarditis as positive a nd as negative for ROCanalysis resulted in an excellent AUC of 0.917. For the m achine learning models, the sensitivity and specificity were as follows: logisti c regression, 0.92 and 0.85; XGBoost, 0.90 and 0.85; decision trees, 0.88 and 0. 86; and ensemble learning, 0.91 and 0.85, respectively. The resulting AUCs were, in the same order: 0.938, 0.937, 0.930, and 0.941, respectively. In this proof- of-concept study, machine learning algorithms achieved improved diagnostic perfo rmance compared to the major/minor weighing system as used in the MDE2015 criter ia. Moreover, these models provide quantifiable certainty levels of the diagnosi s, potentially enhancing interpretability for clinicians. Additionally, they all ow for easy incorporation of new and/or refined criteria, such as the individual weight of advanced imaging modalities such as CTA or [F] FDG PET/CT.”
NETL
NSTL