基于半监督卷积神经网络进行人机不同步的识别

doi:10.3969/j.issn.1006-7795.2022.05.011

首都医科大学学报 ›› 2022, Vol. 43 ›› Issue (5): 734-739.doi: 10.3969/j.issn.1006-7795.2022.05.011

基于半监督卷积神经网络进行人机不同步的识别

周益民¹, 宁泽惺², 罗旭颖^1,, 何璇¹, 杨燕琳¹, 陈光强¹, 李瑞瑞², 周建新^1,3*

1.首都医科大学附属北京天坛医院重症医学科,北京 100070;
2.北京化工大学信息科学与技术学院,北京 100029;
3.神经疾病数字诊疗北京市工程研究中心,北京 100070

收稿日期:2022-08-08 出版日期:2022-10-21 发布日期:2022-10-25
基金资助:
首都市民健康培育(Z161100000116081)。

Detection of patient-ventilator asynchrony based on semi-supervised convolutional neural network

Zhou Yimin¹, Ning Zexing², Luo Xuying¹, He Xuan¹, Yang Yanlin¹, Chen Guangqiang¹, Li Ruirui², Zhou Jianxin^1,3*

1. Department of Critical Care Medicine, Beijing Tiantan Hospital, Capital Medical University, Beijing 100070, China;
2. Beijing University of Chemical Technology,Beijing 100029, China;
3. Beijing Engineering Research Center of Digital Healthcare for Neurological Diseases, Beijing 100070, China

Received:2022-08-08 Online:2022-10-21 Published:2022-10-25
Contact: * E-mail:Zhoujx.cn@icloud.com
Supported by:
Beijing Municipal Science and Technology Commission (Z161100000116081).

摘要/Abstract

摘要： 目的基于半监督卷积神经网络(semi-supervised convolutional neural network, semi-CNN)构建人机不同步现象(patient-ventilator asynchrony, PVA)识别模型,评价其在压力支持通气(pressure support ventilation,PSV)模式下的诊断效能。方法分析85例接受PSV通气脑损伤患者的机械通气数据,结合食道压监测数据进行人工标识。使用Transformer时间序列预测模型对已标识的正常或发生PVA的呼吸进行转化,转化后的数据输入semi-CNN模型判断是否发生PVA。在测试集中验证模型的准确性、灵敏度、特异度以及与专家标识结果的一致性。结果初始训练集包含正常呼吸513次,异常呼吸69次,经过500次迭代后模型收敛。测试集包含正常呼吸48次,异常呼吸24次。在测试集中,Transformer联合semi-CNN模型识别PVA的准确率为0.92(0.83~0.97),灵敏度为0.79(0.58~0.93),特异度为0.98(0.89~1.00),Kappa值为0.80 (0.65~0.95),测试结果与专家人工标识结果具有高度一致性。结论本研究提供了一种基于semi-CNN算法的PVA识别模型,其识别PVA的准确率和特异度高,识别结果与专家人工标识结果的一致性好,可用于临床实时PVA监测。

关键词: 机械通气, 人机不同步, 深度学习, 卷积神经网络, 半监督学习

Abstract: Objective To construct a patient-ventilator asynchrony (PVA) detection model based on semi-supervised convolutional neural networks (semi-CNN)and evaluate its diagnostic efficiency in pressure support ventilation (PSV) mode.Methods The mechanical ventilation data of 85 brain injury patients with PSV mode were analyzed retrospectively. Tracing of flow, and airway pressure combined with esophageal pressure data were used to identify manually the presence of PVA visually. Patient's future breathing observations were predicted by the Transformer-based time series prediction module. The output of the Transformer module was used as the input parameter of the semi-CNN to detect the occurrence of PVA. The accuracy, sensitivity, and specificity of the semi-CNN model and agreement between the semi-CNN model and human experts were validated in the test set. Results The initial training set included 513 normal breaths and 69 PVA breaths. After 500 epochs, the model converges.The test set included 48 normal breaths and 24 PVA breaths. In the test set, the accuracy,the sensitivity, and the specificity of the Transformer + semi-CNN model in identifying PVA are 0.92(0.83－0.97), 0.79(0.58 - 0.93), and 0.98(0.89 －1.00), respectively.There is substantial agreement between the semi-CNN and the human experts in identifying PVA (Kappa=0.80, 95%CI:0.65-0.95, P<0.01). Conclusion This study provided a PVA detecting method based on a semi-CNN algorithm with high accuracy and specificity. The proposed method showed a substantial agreement with human experts in identifying PVA and could be applied in real-time monitoring of PVA clinically.

Key words: mechanical ventilation, patient-ventilator asynchrony, deep learning, convolutional neural network, semi-supervised learning

中图分类号:

周益民, 宁泽惺, 罗旭颖, 何璇, 杨燕琳, 陈光强, 李瑞瑞, 周建新. 基于半监督卷积神经网络进行人机不同步的识别[J]. 首都医科大学学报, 2022, 43(5): 734-739.

Zhou Yimin, Ning Zexing, Luo Xuying, He Xuan, Yang Yanlin, Chen Guangqiang, Li Ruirui, Zhou Jianxin. Detection of patient-ventilator asynchrony based on semi-supervised convolutional neural network[J]. Journal of Capital Medical University, 2022, 43(5): 734-739.

参考文献

[1] Wittenstein J, Huhle R, Leiderman M, et al. Effect of patient-ventilator asynchrony on lung and diaphragmatic injury in experimental acute respiratory distress syndrome in a porcine model[J]. Br J Anaesth,2021,S0007-0912(21)00717-0.
[2] Blanch L, Villagra A, Sales B, et al. Asynchronies during mechanical ventilation are associated with mortality[J]. Intensive Care Med, 2015, 41(4): 633-641.
[3] de Araújo Sousa M L, Magrans R, Hayashi F K, et al. Predictors of asynchronies during assisted ventilation and its impact on clinical outcomes: the EPISYNC cohort study[J]. J Crit Care, 2020, 57: 30-35.
[4] Beitler J R, Sands S A, Loring S H, et al. Quantifying unintended exposure to high tidal volumes from breath stacking dyssynchrony in ARDS: the BREATHE criteria[J]. Intensive Care Med, 2016, 42(9): 1427-1436.
[5] Gilstrap D, MacIntyre N. Patient-ventilator interactions. Implications for clinical management[J]. Am J Respir Crit Care Med, 2013, 188(9): 1058-1068.
[6] Kyo M, Shimatani T, Hosokawa K, et al. Patient-ventilator asynchrony, impact on clinical outcomes and effectiveness of interventions: a systematic review and meta-analysis[J]. J Intensive Care, 2021, 9(1): 50.
[7] Muhamad Sauki N S, Damanhuri N S, Othman N A, et al. Assessing the asynchrony event based on the ventilation mode for mechanically ventilated patients in ICU[J]. Bioengineering (Basel), 2021, 8(12): 222.
[8] Luo X Y, He X, Zhou Y M, et al. Patient-ventilator asynchrony in acute brain-injured patients: a prospective observational study[J]. Ann Intensive Care, 2020, 10(1): 144.
[9] Pham T, Telias I, Beitler J R. Esophageal manometry[J]. Respir Care, 2020, 65(6): 772-792.
[10] Dres M, Rittayamai N, Brochard L. Monitoring patient-ventilator asynchrony[J]. Curr Opin Crit Care, 2016, 22(3): 246-253.
[11] Nilsestuen J O, Hargett K D. Using ventilator graphics to identify patient-ventilator asynchrony[J]. Respir Care, 2005, 50(2): 202-234.
[12] Ramirez I I, Arellano D H, Adasme R S, et al. Ability of ICU health-care professionals to identify patient-ventilator asynchrony using waveform analysis[J]. Respir Care, 2017, 62(2): 144-149.
[13] 刘士远, 萧毅. 基于深度学习的人工智能对医学影像学的挑战和机遇[J]. 中华放射学杂志, 2017, 51(12): 899-901.
[14] 王智杰, 高杰, 孟茜茜, 等. 基于深度学习的人工智能技术在早期胃癌诊断中的应用[J]. 中华消化内镜杂志, 2018, 35(8): 551-556.
[15] Bhinder B, Gilvary C, Madhukar N S, et al. Artificial intelligence in cancer research and precision medicine[J]. Cancer Discov, 2021, 11(4): 900-915.
[16] Gutierrez G. Artificial intelligence in the intensive care unit[J]. Crit Care, 2020, 24(1): 101.
[17] Komorowski M, Celi L A, Badawi O, et al. The artificial intelligence clinician learns optimal treatment strategies for sepsis in intensive care[J]. Nat Med, 2018, 24(11): 1716-1720.
[18] Mulqueeny Q, Redmond S J, Tassaux D, et al. Automated detection of asynchrony in patient-ventilator interaction[J]. Annu Int Conf IEEE Eng Med Biol Soc, 2009, 2009: 5324-5327.
[19] Piquilloud L, Jolliet P, Revelly J P. Automated detection of patient-ventilator asynchrony: new tool or new toy?[J]. Crit Care, 2013, 17(6): 1015.
[20] Nguyen Q T, Pastor D, Lellouche F, et al. Mechanical ventilation system monitoring: automatic detection of dynamic hyperinflation and asynchrony[J]. Annu Int Conf IEEE Eng Med Biol Soc, 2013, 2013: 5207-5210.
[21] Gholami B, Phan T S, Haddad W M, et al. Replicating human expertise of mechanical ventilation waveform analysis in detecting patient-ventilator cycling asynchrony using machine learning[J]. Comput Biol Med, 2018, 97: 137-144.
[22] Blanch L, Sales B, Montanya J, et al. Validation of the Better Care^© system to detect ineffective efforts during expiration in mechanically ventilated patients: a pilot study[J]. Intensive Care Med, 2012, 38(5): 772-780.
[23] Zhang L W, Mao K D, Duan K L, et al. Detection of patient-ventilator asynchrony from mechanical ventilation waveforms using a two-layer long short-term memory neural network[J]. Comput Biol Med, 2020, 120: 103721.
[24] Pan Q, Zhang L W, Jia M Z, et al. An interpretable 1D convolutional neural network for detecting patient-ventilator asynchrony in mechanical ventilation[J]. Comput Methods Programs Biomed, 2021, 204: 106057.
[25] Sottile P D, Albers D, Higgins C, et al. The association between ventilator dyssynchrony, delivered tidal volume, and sedation using a novel automated ventilator dyssynchrony detection algorithm[J]. Crit Care Med, 2018, 46(2): e151-e157.
[26] Rehm G B, Han J, Kuhn B T, et al. Creation of a robust and generalizable machine learning classifier for patient ventilator asynchrony[J]. Methods Inf Med, 2018, 57(4): 208-219.
[27] 孙秀梅, 周建新. 食道压和跨肺压监测[J]. 中华危重病急救医学, 2018, 30(3): 280-283.
[28] 李瑞瑞, 吴桐桐, 宁泽惺, 等. 基于深度学习的机械通气患者人机不同步预警方法[C]//2021中国自动化大会论文集. 北京: 中国自动化学会, 2021: 700-705.
[29] Thille A W, Rodriguez P, Cabello B, et al. Patient-ventilator asynchrony during assisted mechanical ventilation[J]. Intensive Care Med, 2006, 32(10): 1515-1522.
[30] MacIntyre N R, McConnell R, Cheng K C, et al. Patient-ventilator flow dyssynchrony: flow-limited versus pressure-limited breaths[J]. Crit Care Med, 1997, 25(10): 1671-1677.
[31] Holanda M A, Vasconcelos R D S, Ferreira J C, et al. Patient-ventilator asynchrony[J]. J Bras Pneumol, 2018, 44(4): 321-333.

基于半监督卷积神经网络进行人机不同步的识别

Detection of patient-ventilator asynchrony based on semi-supervised convolutional neural network

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 10

编辑推荐

Metrics

本文评价

[1]	张琳琳, 田野, 苗明月, 段雨晴, 田莹, 周建新. 呼吸机自动测量气道阻断压与标准测量方法的一致性[J]. 首都医科大学学报, 2022, 43(5): 728-733.
[2]	程卫, 隆云. 膈肌保护性通气的概念及进展[J]. 首都医科大学学报, 2022, 43(5): 740-746.
[3]	马兰芳, 薛怡蓉. 面向慢性病人群的智能膳食评估系统[J]. 首都医科大学学报, 2021, 42(2): 262-268.
[4]	朱波, 李志强, 席修明. 智能脱机模式对机械通气时间影响的系统评价[J]. 首都医科大学学报, 2013, 34(2): 191-200.
[5]	秦志强;王辰;庞宝森;孙兵;杜敏捷;牛淑杰;龙胜泽. 机械通气对急性肺损伤兔肺水肿影响的实验研究[J]. 首都医科大学学报, 2007, 28(5): 575-578.
[6]	朱光发. 无创通气治疗呼吸衰竭的研究进展[J]. 首都医科大学学报, 2007, 28(1): 47-51.
[7]	王克杰;王力红;魏嘉平;杨强;贾强;冯明;聂秀红;李坤成;尹建国;王立;续大田;张泰昌;王育琴. 机械通气在SARS治疗中的应用[J]. 首都医科大学学报, 2003, 24(4): 410-413.
[8]	杨强;肖汉;张连国;王烁;聂秀红. 8例严重急性呼吸综合征患者有创机械通气治疗总结[J]. 首都医科大学学报, 2003, 24(4): 422-425.
[9]	杨磊;李非;李铎;贾建国;杨鹏;孙家邦. SARS危重患者机械通气后并发症的临床分析及呼吸治疗策略的探讨[J]. 首都医科大学学报, 2003, 24(4): 426-429.
[10]	杨建国;汪俊韬;林秀玉. 56例极重型乙脑患者应用人工呼吸器治疗的并发症和死亡原因分析[J]. 首都医科大学学报, 2001, 22(3): 241-243.