Imputation of the continuous arterial line blood pressure waveform from non-invasive measurements using deep learning - PubMed (original) (raw)

Imputation of the continuous arterial line blood pressure waveform from non-invasive measurements using deep learning

Brian L Hill et al. Sci Rep. 2021.

Abstract

In two-thirds of intensive care unit (ICU) patients and 90% of surgical patients, arterial blood pressure (ABP) is monitored non-invasively but intermittently using a blood pressure cuff. Since even a few minutes of hypotension increases the risk of mortality and morbidity, for the remaining (high-risk) patients ABP is measured continuously using invasive devices, and derived values are extracted from the recorded waveforms. However, since invasive monitoring is associated with major complications (infection, bleeding, thrombosis), the ideal ABP monitor should be both non-invasive and continuous. With large volumes of high-fidelity physiological waveforms, it may be possible today to impute a physiological waveform from other available signals. Currently, the state-of-the-art approaches for ABP imputation only aim at intermittent systolic and diastolic blood pressure imputation, and there is no method that imputes the continuous ABP waveform. Here, we developed a novel approach to impute the continuous ABP waveform non-invasively using two continuously-monitored waveforms that are currently part of the standard-of-care, the electrocardiogram (ECG) and photo-plethysmogram (PPG), by adapting a deep learning architecture designed for image segmentation. Using over 150,000 min of data collected at two separate health systems from 463 patients, we demonstrate that our model provides a highly accurate prediction of the continuous ABP waveform (root mean square error 5.823 (95% CI 5.806-5.840) mmHg), as well as the derived systolic (mean difference 2.398 ± 5.623 mmHg) and diastolic blood pressure (mean difference - 2.497 ± 3.785 mmHg) compared to arterial line measurements. Our approach can potentially be used to measure blood pressure continuously and non-invasively for all patients in the acute care setting, without the need for any additional instrumentation beyond the current standard-of-care.

PubMed Disclaimer

Conflict of interest statement

M.C. is a consultant for Edwards Lifesciences (Irvine, CA) and Masimo Corp (Irvine, CA), and has funded research from Edwards Lifesciences and Masimo Corp. He is also the founder of Sironis and he owns patents and receives royalties for closed loop hemodynamic management technologies that have been licensed to Edwards Lifesciences. I.H. is the founder and President of Extrico health a company that helps hospitals leverage data from their electronic health record for decision making purposes. The software licensed by Extrico Health was used in the extraction of the EHR data in this manuscript. I.H. receives research support and serves as a consultant for Merck.The other authors declare no competing interests concerning this article.

Figures

Figure 1

Examples of input waveforms for 1D V-Net model. (a) 4-s sample of electrocardiogram (ECG) waveform and (b) a 4-s sample photo-plethysmograph (PPG) waveform.

Figure 2

Example ground truth & predicted waveforms. (a) 4-s window (for the input data shown in Fig. 1) and > 3 h records (b,c). The true continuous blood pressure waveform is shown above in green, and the predicted blood pressure waveform shown below in red.

Figure 3

Bland–Altman plots for the MIMIC and UCLA ICU test cohorts. Systolic BP measurements per patient (left), and Diastolic BP measurements per patient (right) using a thirty-two second window; horizontal error bars represent the standard deviation of the blood pressure values, vertical error bars represent the standard deviation of the differences; solid lines indicate the mean difference values, dashed lines indicate the mean difference values +/− 1 and 2 times the standard deviation of the differences. Results for MIMIC are shown in (a), and UCLA in (b).

References

1. Bijker JB, et al. Intraoperative hypotension and perioperative ischemic stroke after general surgery—a nested case–control study. Anesthesiol. J. Am. Soc. Anesthesiol. 2012;116:658–664. -PubMed
1. van Waes JAR, et al. Association between intraoperative hypotension and myocardial injury after vascular surgery. Anesthesiology. 2016;124:35–44. doi: 10.1097/ALN.0000000000000922. -DOI -PubMed
1. Walsh M, et al. Relationship between intraoperative mean arterial pressure and clinical outcomes after noncardiac surgerytoward an empirical definition of hypotension. Anesthesiol. J. Am. Soc. Anesthesiol. 2013;119:507–515. -PubMed
1. Lehman LH, Saeed M, Talmor D, Mark R, Malhotra A. Methods of blood pressure measurement in the ICU. Crit. Care Med. 2013;41:34–40. doi: 10.1097/CCM.0b013e318265ea46. -DOI -PMC -PubMed
1. Maheshwari K, et al. The relationship between ICU hypotension and in-hospital mortality and morbidity in septic patients. Intensive Care Med. 2018;44:857–867. doi: 10.1007/s00134-018-5218-5. -DOI -PMC -PubMed

Imputation of the continuous arterial line blood pressure waveform from non-invasive measurements using deep learning - PubMed (original) (raw)