Pulmonary gas exchange evaluated by machine learning: a computer simulation

Using computer simulation we investigated whether machine learning (ML) analysis of selected ICU monitoring data can quantify pulmonary gas exchange in multi-compartment format. A 21 compartment ventilation/perfusion (V/Q) model of pulmonary blood flow processed 34,551 combinations of cardiac output, hemoglobin concentration, standard P50, base excess, VO₂ and VCO₂ plus three model-defining parameters: shunt, log SD and mean V/Q. From these inputs the model produced paired arterial blood gases, first with the inspired O₂ fraction (FiO₂) adjusted to arterial saturation (SaO₂) = 0.90, and second with FiO₂ increased by 0.1. 'Stacked regressor' ML ensembles were trained/validated on 90% of this dataset. The remainder with shunt, log SD, and mean 'held back' formed the test-set. 'Two-Point' ML estimates of shunt, log SD and mean utilized data from both FiO₂ settings. 'Single-Point' estimates used only data from SaO₂ = 0.90. From 3454 test gas exchange scenarios, two-point shunt, log SD and mean estimates produced linear regression models versus true values with slopes ~ 1.00, intercepts ~ 0.00 and R² ~ 1.00. Kernel density and Bland-Altman plots confirmed close agreement. Single-point estimates were less accurate: R² = 0.77-0.89, slope = 0.991-0.993, intercept = 0.009-0.334. ML applications using blood gas, indirect calorimetry, and cardiac output data can quantify pulmonary gas exchange in terms describing a 20 compartment V/Q model of pulmonary blood flow. High fidelity reports require data from two FiO₂ settings.

Positioning for acute respiratory distress in hospitalised infants and children

BACKGROUND

Acute respiratory distress syndrome (ARDS) is a significant cause of hospitalisation and death in young children. Positioning and mechanical ventilation have been regularly used to reduce respiratory distress and improve oxygenation in hospitalised patients. Due to the association of prone positioning (lying on the abdomen) with sudden infant death syndrome (SIDS) within the first six months, it is recommended that young infants be placed on their back (supine). However, prone positioning may be a non-invasive way of increasing oxygenation in individuals with acute respiratory distress, and offers a more significant survival advantage in those who are mechanically ventilated. There are substantial differences in respiratory mechanics between adults and infants. While the respiratory tract undergoes significant development within the first two years of life, differences in airway physiology between adults and children become less prominent by six to eight years old. However, there is a reduced risk of SIDS during artificial ventilation in hospitalised infants. Thus, an updated review focusing on positioning for infants and young children with ARDS is warranted. This is an update of a review published in 2005, 2009, and 2012.

OBJECTIVES

To compare the effects of different body positions in hospitalised infants and children with acute respiratory distress syndrome aged between four weeks and 16 years.

SEARCH METHODS

We searched CENTRAL, which contains the Acute Respiratory Infections Group's Specialised Register, MEDLINE, Embase, and CINAHL from January 2004 to July 2021.

SELECTION CRITERIA

Randomised controlled trials (RCTs) or quasi-RCTs comparing two or more positions for the management of infants and children hospitalised with ARDS.

DATA COLLECTION AND ANALYSIS

Two review authors independently extracted data from each study. We resolved differences by consensus, or referred to a third contributor to arbitrate. We analysed bivariate outcomes using an odds ratio (OR) and 95% confidence interval (CI). We analysed continuous outcomes using a mean difference (MD) and 95% CI. We used a fixed-effect model, unless heterogeneity was significant (I² statistic > 50%), when we used a random-effects model.

MAIN RESULTS

We included six trials: four cross-over trials, and two parallel randomised trials, with 198 participants aged between 4 weeks and 16 years, all but 15 of whom were mechanically ventilated. Four trials compared prone to supine positions. One trial compared the prone position to good-lung dependent (where the person lies on the side of the healthy lung, e.g. if the right lung was healthy, they were made to lie on the right side), and independent (or non-good-lung independent, where the person lies on the opposite side to the healthy lung, e.g. if the right lung was healthy, they were made to lie on the left side) position. One trial compared good-lung independent to good-lung dependent positions. When the prone (with ventilators) and supine positions were compared, there was no information on episodes of apnoea or mortality due to respiratory events. There was no conclusive result in oxygen saturation (SaO_2; MD 0.40 mmHg, 95% CI -1.22 to 2.66; 1 trial, 30 participants; very low certainty evidence); blood gases, PCO₂ (MD 3.0 mmHg, 95% CI -1.93 to 7.93; 1 trial, 99 participants; low certainty evidence), or PO₂ (MD 2 mmHg, 95% CI -5.29 to 9.29; 1 trial, 99 participants; low certainty evidence); or lung function (PaO₂/FiO₂ ratio; MD 28.16 mmHg, 95% CI -9.92 to 66.24; 2 trials, 121 participants; very low certainty evidence). However, there was an improvement in oxygenation index (FiO₂% X M_PAW/ PaO₂) with prone positioning in both the parallel trials (MD -2.42, 95% CI -3.60 to -1.25; 2 trials, 121 participants; very low certainty evidence), and the cross-over study (MD -8.13, 95% CI -15.01 to -1.25; 1 study, 20 participants). Derived indices of respiratory mechanics, such as tidal volume, respiratory rate, and positive end-expiratory pressure (PEEP) were reported. There was an apparent decrease in tidal volume between prone and supine groups in a parallel study (MD -0.60, 95% CI -1.05 to -0.15; 1 study, 84 participants; very low certainty evidence). When prone and supine positions were compared in a cross-over study, there were no conclusive results in respiratory compliance (MD 0.07, 95% CI -0.10 to 0.24; 1 study, 10 participants); changes in PEEP (MD -0.70 cm H₂O, 95% CI -2.72 to 1.32; 1 study, 10 participants); or resistance (MD -0.00, 95% CI -0.05 to 0.04; 1 study, 10 participants). One study reported adverse events. There were no conclusive results for potential harm between groups in extubation (OR 0.57, 95% CI 0.13 to 2.54; 1 trial, 102 participants; very low certainty evidence); obstructions of the endotracheal tube (OR 5.20, 95% CI 0.24 to 111.09; 1 trial, 102 participants; very low certainty evidence); pressure ulcers (OR 1.00, 95% CI 0.41 to 2.44; 1 trial, 102 participants; very low certainty evidence); and hypercapnia (high levels of arterial carbon dioxide; OR 3.06, 95% CI 0.12 to 76.88; 1 trial, 102 participants; very low certainty evidence). One study (50 participants) compared supine positions to good-lung dependent and independent positions. There was no conclusive evidence that PaO₂ was different between supine and good-lung dependent positioning (MD 3.44 mm Hg, 95% CI -23.12 to 30.00; 1 trial, 25 participants; very low certainty evidence). There was also no conclusive evidence for supine position and good-lung independent positioning (MD -2.78 mmHg, 95% CI -28.84, 23.28; 25 participants; very low certainty evidence); or between good-lung dependent and independent positioning (MD 6.22, 95% CI -21.25 to 33.69; 1 trial, 25 participants; very low certainty evidence). As most trials did not describe how possible biases were addressed, the potential for bias in these findings is unclear.

AUTHORS' CONCLUSIONS

Although included studies suggest that prone positioning may offer some advantage, there was little evidence to make definitive recommendations. There appears to be low certainty evidence that positioning improves oxygenation in mechanically ventilated children with ARDS. Due to the increased risk of SIDS with prone positioning and lung injury with artificial ventilation, it is recommended that hospitalised infants and children should only be placed in this position while under continuous cardiorespiratory monitoring.

Correlation of Oxygen Index, Oxygen Saturation Index, and PaO2/FiO2 Ratio in Invasive Mechanically Ventilated Adults

Background

With the oxygen saturation index (OSI) being a noninvasive surrogate for oxygen index (OI) and P/F ratio, examining the correlation between PaO₂/FiO₂ (P/F ratio), OI, and OSI in mechanically ventilated adults will benefit in those settings where arterial blood gas monitoring is not readily accessible.

Materials and methods

Data were collected for patients ≥18 years who were under invasive (endotracheal intubation) mechanical ventilation at medical or surgical wards in a tertiary care hospital.

Results

After natural log transformation, the correlations between P/F ratio and OI (r = −0.94) and OI and OSI (r = 0.82) were strong, but weaker between P/F ratio and OSI (r = −0.69).

Conclusion

Future bigger studies are needed to evaluate whether monitoring OSI and/or OI over P/F ratio will impact treatment outcomes.

How to cite this article

Vadi S. Correlation of Oxygen Index, Oxygen Saturation Index, and PaO₂/FiO₂ Ratio in Invasive Mechanically Ventilated Adults. Indian J Crit Care Med 2021;25(1):54–55.

Post-extubation continuous positive airway pressure improves oxygenation after pediatric laparoscopic surgery: A randomized controlled trial

BACKGROUND

Effects of intraoperative recruitment maneuvers (RMs) on oxygenation and pulmonary compliance are lost during recovery if high inspired oxygen and airway suctioning are used. We investigated the effect of post-extubation noninvasive CPAP mask application on the alveolar arterial oxygen difference [(A-a) DO₂ ] after pediatric laparoscopic surgery.

METHODS

Sixty patients (1-6 years) were randomly allocated to three groups of 20 patients, to receive zero end-expiratory pressure (ZEEP group), RM with decremental PEEP titration only (RM group), or followed with post-extubation CPAP for 5 minutes (RM-CPAP group). Primary outcome was [(A-a) DO₂ ] at 1 hour postoperatively. Secondary outcomes were respiratory mechanics, arterial blood gas analysis, hemodynamics, and adverse events.

RESULTS

At 1 hour postoperatively, mean [(A-a) DO₂ ] (mm Hg) was lower in the RM-CPAP group (41.5 ± 13.2, [95% CI 37.6-45.8]) compared to (80.2 ± 13.7 [72.6-87.5], P < 0.0001] and (59.2 ± 14.6, [54.8-62.6], P < 0.001) in the ZEEP and RM groups. The mean PaO₂ (mm Hg) at 1 hour postoperatively was higher in the RM-CPAP group (156.2 ± 18.3 [95% CI 147.6-164.7]) compared with the ZEEP (95.9 ± 15.9 [88.5-103.3], P < 0.0001) and RM groups (129.1 ± 15.9 [121.6-136.5], P < 0.0001). At 12 hours postoperatively, mean [(A-a) DO₂ ] and PaO₂ were (9.6 ± 2.1 [8.4-10.8]) and (91.9 ± 9.4 [87.5-96.3]) in the RM-CPAP group compared to (25.8 ± 5.5 [23.6-27.6]) and (69.9 ± 5.5 [67.4-72.5], P < 0.0001) in the ZEEP group and (34.3 ± 13.2, [28.4-40.2], P < 0.0001) and (74.03 ± 9.8 [69.5-78.6], P < 0.0001) in the RM group. No significant differences of perioperative adverse effects were found between groups.

CONCLUSIONS

An RM done after pneumoperitoneum inflation followed by decremental PEEP titration improved oxygenation at 1 hour postoperatively. The addition of an early post-extubation noninvasive CPAP mask ventilation improved oxygenation at 12 hours postoperatively.

ETORPHINE-KETAMINE-MEDETOMIDINE TOTAL INTRAVENOUS ANESTHESIA IN WILD IMPALA (AEPYCEROS MELAMPUS) OF 120-MINUTE DURATION

There is a growing necessity to perform long-term anesthesia in wildlife, especially antelope. The costs and logistics of transporting wildlife to veterinary practices make surgical intervention a high-stakes operation. Thus there is a need for a field-ready total intravenous anesthesia (TIVA) infusion to maintain anesthesia in antelope. This study explored the feasibility of an etorphine-ketamine-medetomidine TIVA for field anesthesia. Ten wild-caught, adult impala ( Aepyceros melampus ) were enrolled in the study. Impala were immobilized with a standardized combination of etorphine (2 mg) and medetomidine (2.2 mg), which equated to a median (interquartile range [IQR]) etorphine and medetomidine dose of 50.1 (46.2-50.3) and 55.1 (50.8-55.4) μg/kg, respectively. Recumbency was attained in a median (IQR) time of 13.9 (12.0-16.5) min. Respiratory gas tensions, spirometry, and arterial blood gas were analyzed over a 120-min infusion. Once instrumented, the TIVA was infused as follows: etorphine at a variable rate initiated at 40 μg/kg per hour (adjusted according to intermittent deep-pain testing); ketamine and medetomidine at a fixed rate of 1.5 mg/kg per hour and 5 μg/kg per hour, respectively. The etorphine had an erratic titration to clinical effect in four impala. Arterial blood pressure and respiratory and heart rates were all within normal physiological ranges. However, arterial blood gas analysis revealed severe hypoxemia, hypercapnia, and acidosis. Oxygenation and ventilation indices were calculated and highlighted possible co-etiologies to the suspected etorphine-induced respiratory depression as the cause of the blood gas derangements. Impala recovered in the boma post atipamezole (13 mg) and naltrexone (42 mg) antagonism of medetomidine and etorphine, respectively. The etorphine-ketamine-medetomidine TIVA protocol for impala may be sufficient for field procedures of up to 120-min duration. However, hypoxemia and hypercapnia are of paramount concern and thus oxygen supplementation should be considered mandatory. Other TIVA combinations may be superior and warrant further investigation.

Can computer simulators accurately represent the pathophysiology of individual COPD patients?

BACKGROUND

Computer simulation models could play a key role in developing novel therapeutic strategies for patients with chronic obstructive pulmonary disease (COPD) if they can be shown to accurately represent the pathophysiological characteristics of individual patients.

METHODS

We evaluated the capability of a computational simulator to reproduce the heterogeneous effects of COPD on alveolar mechanics as captured in a number of different patient datasets.

RESULTS

Our results show that accurately representing the pathophysiology of individual COPD patients necessitates the use of simulation models with large numbers (up to 200) of compartments for gas exchange. The tuning of such complex simulation models 'by hand' to match patient data is not feasible, and thus we present an automated approach based on the use of global optimization algorithms and high-performance computing. Using this approach, we are able to achieve extremely close matches between the simulator and a range of patient data including PaO2, PaCO2, pulmonary deadspace fraction, pulmonary shunt fraction, and ventilation/perfusion (/Q) curves. Using the simulator, we computed combinations of ventilator settings that optimally manage the trade-off between ensuring adequate gas exchange and minimizing the risk of ventilator-associated lung injury for an individual COPD patient.

CONCLUSIONS

Our results significantly strengthen the credibility of computer simulation models as research tools for the development of novel management protocols in COPD and other pulmonary disease states.

Optimization of mechanical ventilator settings for pulmonary disease states

The selection of mechanical ventilator settings that ensure adequate oxygenation and carbon dioxide clearance while minimizing the risk of ventilator-associated lung injury (VALI) is a significant challenge for intensive-care clinicians. Current guidelines are largely based on previous experience combined with recommendations from a limited number of in vivo studies whose data are typically more applicable to populations than to individuals suffering from particular diseases of the lung. By combining validated computational models of pulmonary pathophysiology with global optimization algorithms, we generate in silico experiments to examine current practice and uncover optimal combinations of ventilator settings for individual patient and disease states. Formulating the problem as a multiobjective, multivariable constrained optimization problem, we compute settings of tidal volume, ventilation rate, inspiratory/expiratory ratio, positive end-expiratory pressure and inspired fraction of oxygen that optimally manage the tradeoffs between ensuring adequate oxygenation and carbon dioxide clearance and minimizing the risk of VALI for different pulmonary disease scenarios.

Prediction of arterial oxygen partial pressure after changes in FIO₂: validation and clinical application of a novel formula

BACKGROUND

Existing methods allow prediction of Pa(O₂) during adjustment of Fi(O₂). However, these are cumbersome and lack sufficient accuracy for use in the clinical setting. The present studies aim to extend the validity of a novel formula designed to predict Pa(O₂) during adjustment of Fi(O₂) and to compare it with the current methods.

METHODS

Sixty-seven new data sets were collected from 46 randomly selected, mechanically ventilated patients. Each data set consisted of two subsets (before and 20 min after Fi(O₂) adjustment) and contained ventilator settings, pH, and arterial blood gas values. We compared the accuracy of Pa(O₂) prediction using a new formula (which utilizes only the pre-adjustment Pa(O₂) and pre- and post-adjustment Fi(O₂) with prediction using assumptions of constant Pa(O₂)/Fi(O₂) or constant Pa(O₂)/Pa(O₂). Subsequently, 20 clinicians predicted Pa(O₂) using the new formula and using Nunn's isoshunt diagram. The accuracy of the clinician's predictions was examined.

RESULTS

The 95% limits of agreement (LA(95%)) between predicted and measured Pa(O₂) in the patient group were: new formula 0.11 (2.0) kPa, Pa(O₂)/Fi(O₂) -1.9 (4.4) kPa, and Pa(O₂)/Pa(O₂) -1.0 (3.6) kPa. The LA(95%) of clinicians' predictions of Pa(O₂) were 0.56 (3.6) kPa (new formula) and -2.7 (6.4) kPa (isoshunt diagram).

CONCLUSIONS

The new formula's prediction of changes in Pa(O₂) is acceptably accurate and reliable and better than any other existing method. Its use by clinicians appears to improve accuracy over the most popular existing method. The simplicity of the new method may allow its regular use in the critical care setting.

Assessing gas exchange in acute lung injury/acute respiratory distress syndrome: diagnostic techniques and prognostic relevance

PURPOSE OF REVIEW

To provide the most recent insights on the assessment of gas exchange in acute lung injury.

RECENT FINDINGS

Central venous blood may be used as a surrogate of arterial blood to assess carbon dioxide tension and acid-base status. In contrast arterial oxygenation cannot be estimated with confidence from venous blood. However, the use of venous blood associated with pulse oximetry may provide the SvO2 which is useful for monitoring and targeting the resuscitation therapy. Impaired CO2 clearance and increased dead space have been confirmed as useful prognostic indices of structural lung damage and mortality in acute respiratory failure. A simplified technique based on multiple inert gas technique has been described to assess ventilation-perfusion mismatch while a new analysis of pulse oximetry has been suggested to detect lung opening and closing. Finally, new insight has been provided on the relationship between lung anatomy, as detected by computed tomography, oxygenation and CO2 clearance.

SUMMARY

Although oxygenation assessment is of primary importance during respiratory lung injury, dead space and CO2 retention are more strictly associated with outcome. The association of central venous blood analysis and pulse oximetry may provide more information than arterial blood alone.

Computational modelling of lung injury: is there potential for benefit?

State-of-the-art medical care of the victims of current conflicts is generating large quantities of quality clinical data as a by-product. Observational research based on these data is beginning to have a profound influence on the clinical management of both military and civilian trauma patients. Computational modelling based on these datasets may offer the ability to investigate clinical treatment strategies that are practically, ethically or scientifically impossible to investigate on the front line. This article reviews the potential of this novel technology to aid development of treatment for blast lung and other unresolved medical scenarios.