Estimated Pao 2: A Continuous and Noninvasive Method to Estimate Pao 2 and Oxygenation Index

Sauthier, Michaël; Tuli, Gaurav; Jouvet, Philippe; Brownstein, John S.; Randolph, Adrienne G.

doi:10.1097/cce.0000000000000546

Cited by 17 publications

(11 citation statements)

References 40 publications

(62 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We showed that machine learning models outperformed previously published equations in terms of imputing PaO 2 from SpO 2 in the mechanically-ventilated adult population. Consistent with our findings, Sauthier et al, utilized neural network models to validate a continuous and noninvasive method of hypoxemia estimation in pediatric population 28 . They utilized convolutional neural network (CNN), long short-term memory network (LSTM), and multilayer perceptron (MLP) to impute PaO 2 .…”

Section: Discussionsupporting

confidence: 88%

Machine learning based algorithms to impute PaO2 from SpO2 values and development of an online calculator

Ren

Zupetic

Tabary

et al. 2022

Sci Rep

View full text Add to dashboard Cite

We created an online calculator using machine learning (ML) algorithms to impute the partial pressure of oxygen (PaO2)/fraction of delivered oxygen (FiO2) ratio using the non-invasive peripheral saturation of oxygen (SpO2) and compared the accuracy of the ML models we developed to published equations. We generated three ML algorithms (neural network, regression, and kernel-based methods) using seven clinical variable features (N = 9900 ICU events) and subsequently three features (N = 20,198 ICU events) as input into the models. Data from mechanically ventilated ICU patients were obtained from the publicly available Medical Information Mart for Intensive Care (MIMIC III) database and used for analysis. Compared to seven features, three features (SpO2, FiO2 and PEEP) were sufficient to impute PaO2 from the SpO2. Any of the ML models enabled imputation of PaO2 from the SpO2 with lower error and showed greater accuracy in predicting PaO2/FiO2 ≤ 150 compared to the previously published log-linear and non-linear equations. To address potential hidden hypoxemia that occurs more frequently in Black patients, we conducted sensitivity analysis and show ML models outperformed published equations in both Black and White patients. Imputation using data from an independent validation cohort of ICU patients (N = 133) showed greater accuracy with ML models.

show abstract

Section: Discussionsupporting

confidence: 88%

Machine learning based algorithms to impute PaO2 from SpO2 values and development of an online calculator

Ren

Zupetic

Tabary

et al. 2022

Sci Rep

View full text Add to dashboard Cite

show abstract

“…This frequency of Spo 2 measurements greater than 97% is consistent with data from many other ICU cohorts (20)(21)(22)(23). We thus used the estimated Pao 2 /Fio 2 ratio (ePF ratio), as described by Sauthier et al (24) to overcome this problem. For every available data point, ePao 2 was estimated from the Spo 2 using equation by Sauthier et al (24) valid across all values of Spo 2 .…”

Section: Research In Contextsupporting

confidence: 73%

“…This frequency of Sp o 2 measurements greater than 97% is consistent with data from many other ICU cohorts (20–23). We thus used the estimated Pa o 2 /F io 2 ratio (ePF ratio), as described by Sauthier et al (24) to overcome this problem. For every available data point, ePa o 2 was estimated from the Sp o 2 using equation by Sauthier et al (24) valid across all values of Sp o 2 .…”

Section: Methodsmentioning

confidence: 99%

Elevated Diaphragmatic Tonic Activity in PICU Patients: Age-Specific Definitions, Prevalence, and Associations*

Plante

Poirier

Guay

et al. 2023

Pediatric Critical Care Medicine

View full text Add to dashboard Cite

OBJECTIVES: Tonic diaphragmatic activity (tonic Edi, i.e., sustained diaphragm activation throughout expiration) reflects diaphragmatic effort to defend end-expiratory lung volumes. Detection of such elevated tonic Edi may be useful in identifying patients who need increased positive end-expiratory pressure. We aimed to: 1) identify age-specific definitions for elevated tonic Edi in ventilated PICU patients and 2) describe the prevalence and factors associated with sustained episodes of high tonic Edi. DESIGN: Retrospective study using a high-resolution database. SETTING: Single-center tertiary PICU. PATIENTS: Four hundred thirty-one children admitted between 2015 and 2020 with continuous Edi monitoring. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We characterized our definition of tonic Edi using data from the recovery phase of respiratory illness (i.e., final 3 hr of Edi monitoring, excluding patients with significant persistent disease or with diaphragm pathology). High tonic Edi was defined as population data exceeding the 97.5th percentile, which for infants younger than 1 year was greater than 3.2 μV and for older children as greater than 1.9 μV. These thresholds were then used to identify patients with episodes of sustained elevated tonic Edi in the first 48 hours of ventilation (acute phase). Overall, 62 of 200 (31%) of intubated patients and 138 of 222 (62%) of patients on noninvasive ventilation (NIV) had at least one episode of high tonic Edi. These episodes were independently associated with the diagnosis of bronchiolitis (intubated patients: adjusted odds [aOR], 2.79 [95% CI, 1.12–7.11]); NIV patients: aOR, 2.71 [1.24–6.0]). There was also an association with tachypnea and, in NIV patients, more severe hypoxemia. CONCLUSIONS: Our proposed definition of elevated tonic Edi quantifies abnormal diaphragmatic activity during expiration. Such a definition may help clinicians to identify those patients using abnormal effort to defend end-expiratory lung volume. In our experience, high tonic Edi episodes are frequent, especially during NIV and in patients with bronchiolitis.

show abstract

“…Ventilator data were obtained through a continuous data feed to the monitor, and when this data feed was not set up or disconnected, some variables were not available. When a capillary blood gas was used, PaO 2 was estimated from SpO 2 , using the reported formula by Sauthier et al to calculate OI and PF ratio [18]. with the closest blood gas to the time point of interest, which was within the time window, identified for the analysis.…”

Section: Variable Definitionmentioning

confidence: 99%

The end-tidal alveolar dead space fraction for risk stratification during the first week of invasive mechanical ventilation: an observational cohort study

et al. 2023

View full text Add to dashboard Cite

Background The end-tidal alveolar dead space fraction (AVDSf = [PaCO2−PETCO2]/PaCO2) is a metric used to estimate alveolar dead space. Higher AVDSf on the first day of mechanical ventilation is associated with mortality and fewer ventilator-free days. It is not clear if AVDSf is associated with length of ventilation in survivors, how AVDSf performs for risk stratification beyond the first day of ventilation, or whether AVDSf adds predictive value to oxygenation (oxygenation index [OI]) or severity of illness (Pediatric Risk of Mortality [PRISM III]) markers. Methods Retrospective single-center observational cohort study of children and young adults receiving invasive mechanical ventilation. In those with arterial or capillary blood gases, AVDSf was calculated at the time of every blood gas for the first week of mechanical ventilation. Results There were 2335 children and young adults (median age 5.8 years [IQR 1.2, 13.2]) enrolled with 8004 analyzed AVDSf values. Higher AVDSf was associated with mortality and longer length of ventilation in survivors throughout the first week of ventilation after controlling for OI and PRISM III. Higher OI was not associated with increased mortality until ≥ 48 h of ventilation after controlling for AVDSf and PRISM III. When using standardized variables, AVDSf effect estimates were generally higher than OI for mortality, whereas OI effect estimates were generally higher than AVDSf for the length of ventilation in survivors. An AVDSf > 0.3 was associated with a higher mortality than an AVDSf < 0.2 within each pediatric acute respiratory distress syndrome severity category. The maximum AVDSf within 12 h of intensive care unit admission demonstrated good risk stratification for mortality (AUC 0.768 [95% CI 0.732, 0.803]). AVDSf did not improve mortality risk stratification when added to PRISM III but did improve mortality risk stratification when added to the gas exchange components of PRISM III (minimum 12-h PaO2 and maximum 12-h PCO2) (p < 0.00001). Conclusions AVDSf is associated with mortality and length of ventilation in survivors throughout the first week of invasive mechanical ventilation. Some analyses suggest AVDSf may better stratify mortality risk than OI, whereas OI may better stratify risk for prolonged ventilation in survivors than AVDSf.

show abstract

Estimated Pao 2: A Continuous and Noninvasive Method to Estimate Pao 2 and Oxygenation Index

Abstract: Supplemental Digital Content is available in the text.

Cited by 17 publications

References 40 publications

Machine learning based algorithms to impute PaO2 from SpO2 values and development of an online calculator

Machine learning based algorithms to impute PaO2 from SpO2 values and development of an online calculator

Elevated Diaphragmatic Tonic Activity in PICU Patients: Age-Specific Definitions, Prevalence, and Associations*

The end-tidal alveolar dead space fraction for risk stratification during the first week of invasive mechanical ventilation: an observational cohort study

Contact Info

Product

Resources

About