Department of Nutrition and Exercise Physiology, Washington State University-Spokane Health Sciences, Elson S. Floyd College of Medicine, 412 E. Spokane Falls Blvd., Spokane, WA, 99202-2131, USA.
Department of Medical Education and Clinical Sciences, Washington State University-Spokane Health Sciences, Elson S. Floyd College of Medicine, Spokane, WA, USA.
Sci Rep. 2023 Oct 11;13(1):17247. doi: 10.1038/s41598-023-44331-z.
Identification of ventilatory constraint is a key objective of clinical exercise testing. Expiratory flow-limitation (EFL) is a well-known type of ventilatory constraint. However, EFL is difficult to measure, and commercial metabolic carts do not readily identify or quantify EFL. Deep machine learning might provide a new approach for identifying EFL. The objective of this study was to determine if a convolutional neural network (CNN) could accurately identify EFL during exercise in adults in whom baseline airway function varied from normal to mildly obstructed. 2931 spontaneous exercise flow-volume loops (eFVL) were placed within the baseline maximal expiratory flow-volume curves (MEFV) from 22 adults (15 M, 7 F; age, 32 yrs) in whom lung function varied from normal to mildly obstructed. Each eFVL was coded as EFL or non-EFL, where EFL was defined by eFVLs with expired airflow meeting or exceeding the MEFV curve. A CNN with seven hidden layers and a 2-neuron softmax output layer was used to analyze the eFVLs. Three separate analyses were conducted: (1) all subjects (n = 2931 eFVLs, [GR]), (2) subjects with normal spirometry (n = 1921 eFVLs [GR]), (3) subjects with mild airway obstruction (n = 1010 eFVLs, [GR]). The final output of the CNN was the probability of EFL or non-EFL in each eFVL, which is considered EFL if the probability exceeds 0.5 or 50%. Baseline forced expiratory volume in 1 s/forced vital capacity was 0.77 (94% predicted) in GR, 0.83 (100% predicted) in GR, and 0.69 (83% predicted) in GR. CNN model accuracy was 90.6, 90.5, and 88.0% in GR, GR and GR, respectively. Negative predictive value (NPV) was higher than positive predictive value (PPV) in GR (93.5 vs. 78.2% for NPV vs. PPV). In GR, PPV was slightly higher than NPV (89.5 vs. 84.5% for PPV vs. NPV). A CNN performed very well at identifying eFVLs with EFL during exercise. These findings suggest that deep machine learning could become a viable tool for identifying ventilatory constraint during clinical exercise testing.
识别通气受限是临床运动测试的一个关键目标。呼气流量受限（EFL）是一种众所周知的通气受限类型。然而，EFL 很难测量，而且商业代谢车也不容易识别或量化 EFL。深度学习可能为识别 EFL 提供一种新方法。本研究的目的是确定卷积神经网络（CNN）是否可以在基线气道功能从正常到轻度阻塞的成年人中准确识别运动时的 EFL。从 22 名成年人（15 名男性，7 名女性；年龄 32 岁）的基线最大呼气流量-容积曲线（MEFV）中放置了 2931 个自发运动流量-容积环（eFVL），这些成年人的肺功能从正常到轻度阻塞不等。每个 eFVL 都被编码为 EFL 或非 EFL，其中 EFL 定义为呼气流量达到或超过 MEFV 曲线的 eFVL。具有七个隐藏层和两个神经元 softmax 输出层的 CNN 用于分析 eFVL。进行了三项独立分析：（1）所有受试者（n=2931 个 eFVL，[GR]），（2）肺功能正常的受试者（n=1921 个 eFVL，[GR]），（3）轻度气道阻塞的受试者（n=1010 个 eFVL，[GR]）。CNN 的最终输出是每个 eFVL 中 EFL 或非 EFL 的概率，如果概率超过 0.5 或 50%，则认为是 EFL。GR 中的基线 1 秒用力呼气量/用力肺活量为 0.77（预测值的 94%），GR 中的 0.83（预测值的 100%），GR 中的 0.69（预测值的 83%）。在 GR、GR 和 GR 中，CNN 模型的准确性分别为 90.6%、90.5%和 88.0%。在 GR 中，阴性预测值（NPV）高于阳性预测值（PPV）（NPV 为 93.5%，PPV 为 78.2%）。在 GR 中，PPV 略高于 NPV（PPV 为 89.5%，NPV 为 84.5%）。CNN 在识别运动时 EFL 的 eFVL 方面表现非常出色。这些发现表明，深度学习可能成为识别临床运动测试中通气受限的一种可行工具。
Zotero 插件
