Automated analysis of respiratory behavior in extremely preterm infants and extubation readiness

Using machine-learning models to predict extubation failure in neonates with bronchopulmonary dysplasia

AIM

To develop a decision-support tool for predicting extubation failure (EF) in neonates with bronchopulmonary dysplasia (BPD) using a set of machine-learning algorithms.

METHODS

A dataset of 284 BPD neonates on mechanical ventilation was used to develop predictive models via machine-learning algorithms, including extreme gradient boosting (XGBoost), random forest, support vector machine, naïve Bayes, logistic regression, and k-nearest neighbor. The top three models were assessed by the area under the receiver operating characteristic curve (AUC), and their performance was tested by decision curve analysis (DCA). Confusion matrix was used to show the high performance of the best model. The importance matrix plot and SHapley Additive exPlanations values were calculated to evaluate the feature importance and visualize the results. The nomogram and clinical impact curves were used to validate the final model.

RESULTS

According to the AUC values and DCA results, the XGboost model performed best (AUC = 0.873, sensitivity = 0.896, specificity = 0.838). The nomogram and clinical impact curve verified that the XGBoost model possessed a significant predictive value. The following were predictive factors for EF: pO2, hemoglobin, mechanical ventilation (MV) rate, pH, Apgar score at 5 min, FiO2, C-reactive protein, Apgar score at 1 min, red blood cell count, PIP, gestational age, highest FiO2 at the first 24 h, heart rate, birth weight, pCO2. Further, pO2, hemoglobin, and MV rate were the three most important factors for predicting EF.

CONCLUSIONS

The present study indicated that the XGBoost model was significant in predicting EF in BPD neonates with mechanical ventilation, which is helpful in determining the right extubation time among neonates with BPD to reduce the occurrence of complications.

Peri-extubation settings in preterm neonates: a systematic review and meta-analysis

OBJECTIVE

To systematically review: 1) peri-extubation settings; and 2) association between peri-extubation settings and outcomes in preterm neonates.

STUDY DESIGN

In this systematic review, studies were eligible if they reported patient-data on peri-extubation settings (objective 1) and/or evaluated peri-extubation levels in relation to clinical outcomes (objective 2). Data were meta-analyzed when appropriate using random-effects model.

RESULTS

Of 9681 titles, 376 full-texts were reviewed and 101 included. The pooled means of peri-extubation settings were summarized. For objective 2, three experimental studies were identified comparing post-extubation CPAP levels. Meta-analyses revealed lower odds for treatment failure [pooled OR 0.46 (95% CI 0.27-0.76); 3 studies, 255 participants] but not for re-intubation [pooled OR 0.66 (0.22-1.97); 3 studies, 255 participants] with higher vs. lower CPAP.

CONCLUSIONS

Summary of peri-extubation settings may guide clinicians in their own practices. Higher CPAP levels may reduce extubation failure, but more data on peri-extubation settings that optimize outcomes are needed.

Automated prediction of extubation success in extremely preterm infants: the APEX multicenter study

BACKGROUND

Extremely preterm infants are frequently subjected to mechanical ventilation. Current prediction tools of extubation success lacks accuracy.

METHODS

Multicenter study including infants with birth weight ≤1250 g undergoing their first extubation attempt. Clinical data and cardiorespiratory signals were acquired before extubation. Primary outcome was prediction of extubation success. Automated analysis of cardiorespiratory signals, development of clinical and cardiorespiratory features, and a 2-stage Clinical Decision-Balanced Random Forest classifier were used. A leave-one-out cross-validation was done. Performance was analyzed by ROC curves and determined by balanced accuracy. An exploratory analysis was performed for extubations before 7 days of age.

RESULTS

A total of 241 infants were included and 44 failed (18%) extubation. The classifier had a balanced accuracy of 73% (sensitivity 70% [95% CI: 63%, 76%], specificity 75% [95% CI: 62%, 88%]). As an additional clinical-decision tool, the classifier would have led to an increase in extubation success from 82% to 93% but misclassified 60 infants who would have been successfully extubated. In infants extubated before 7 days of age, the classifier identified 16/18 failures (specificity 89%) and 73/105 infants with success (sensitivity 70%).

CONCLUSIONS

Machine learning algorithms may improve a balanced prediction of extubation outcomes, but further refinement and validation is required.

IMPACT

A machine learning-derived predictive model combining clinical data with automated analyses of individual cardiorespiratory signals may improve the prediction of successful extubation and identify infants at higher risk of failure with a good balanced accuracy. Such multidisciplinary approach including medicine, biomedical engineering and computer science is a step forward as current tools investigated to predict extubation outcomes lack sufficient balanced accuracy to justify their use in future trials or clinical practice. Thus, this individualized assessment can optimize patient selection for future trials of extubation readiness by decreasing exposure of low-risk infants to interventions and maximize the benefits of those at high risk.

Different Settings of Nonsynchronized Bilevel Nasal Continuous Positive Airway Pressure and Respiratory Function in Preterm Infants: A Pilot Study

OBJECTIVE

With this study, we evaluated the short-term effects of different modes and settings of noninvasive respiratory support on gas exchange, breathing parameters, and thoracoabdominal synchrony in preterm infants in the acute phase of moderate respiratory distress syndrome.

STUDY DESIGN

A feasibility crossover trial was conducted in neonates < 32 weeks' gestation on nasal continuous positive airway pressure (n-CPAP) or bilevel n-CPAP. Infants were delivered the following settings in consecutive order for 10 minutes each: • n-CPAP (5 cm H₂O) • bilevel n-CPAP 1 (Pres low = 5 cm H₂O, Pres high = 7 cm H₂O, T-high = 1 second, rate = 30/min) • n-CPAP (5 cm H₂O) • bilevel n-CPAP 2 (Pres low = 5 cm H₂O, Pres high = 7 cm H₂O, T-high = 2 second, rate = 15/min) • n-CPAP (5 cm H₂O). During each phase, physiologic parameters were recorded; the thoracoabdominal synchrony expressed by the phase angle (Φ) and other respiratory patterns were monitored by noncalibrated respiratory inductance plethysmography.

RESULTS

Fourteen preterm infants were analyzed. The mean CPAP level was significantly lower in the n-CPAP period compared with bilevel n-CPAP 1 and 2 (p = 0.03). Higher values were achieved with bilevel n-CPAP 2 (6.2 ± 0.6 vs. 5.7 ± 0.5 cm H₂O, respectively; p < 0.05). No statistical difference in the Φ was detected, nor between the three settings.

CONCLUSION

Our study did not show any superiority of bilevel n-CPAP over n-CPAP. However, nonsynchronized bilevel n-CPAP might be helpful when additional pressure is needed.

KEY POINTS

· There is currently a high degree of uncertainty about the superiority of one modality and setting of noninvasive respiratory over another.. · Our study confirmed that non-synchronized bilevel n-CPAP might be helpful when additional pressure is needed for recruitment.. · A T-high of 1 second could possibly be better tolerated in this population, but further research is needed..

Can Fraction of Inspired Oxygen Predict Extubation Failure in Preterm Infants?

Background: Prolonged mechanical ventilation in preterm infants may cause complications. We aimed to analyze the variables affecting extubation outcomes in preterm infants at high risk of extubation failure. Methods: This was a single-center, observational, retrospective study. Extubation failure was defined as survival with the need for reintubation within 72 h. Successfully extubated neonates (group 1) were compared to those with failed extubation (group 2). Multivariate logistic regression analysis evaluated factors that predicted extubation outcomes. Results: Eighty infants with a birth weight under 1000 g and/or gestational age (GA) under 28 weeks were included. Extubation failure occurred in 29 (36.2%) and success in 51 (63.8%) neonates. Most failures (75.9%) occurred within 24 h. Pre-extubation inspired oxygen fraction (FiO₂) of 27% had a sensitivity of 58.6% and specificity of 64.7% for extubation failure. Post-extubation FiO₂ of 32% had a sensitivity of 65.5% and specificity of 62.8% for failure. Prolonged membrane rupture (PROM) and high GA were associated with extubation success in multivariate logistic regression analysis. Conclusions: High GA and PROM were associated with extubation success. Pre- and post-extubation FiO₂ values were not significantly predictive of extubation failure. Further studies should evaluate if overall assessment, including ventilatory parameters and clinical factors, can predict extubation success in neonates.

Decision to extubate extremely preterm infants: art, science or gamble?

In the modern era of neonatology, mechanical ventilation has been restricted to a smaller and more immature population of extremely preterm infants. Given the adverse outcomes associated with mechanical ventilation, every effort is made to extubate these infants as early as possible. However, the scientific basis for determining extubation readiness remains imprecise and primarily guided by clinical judgement, which is highly variable and subjective. In the absence of accurate tools to assess extubation readiness, many infants fail their extubation attempt and require reintubation, which also increases complications. Recent advances in the field have led to unravelling some of the complexities surrounding extubation in this population. This review aims to synthesise the available knowledge and provide a more evidence-based approach towards the reporting of extubation outcomes and assessment of extubation readiness in extremely preterm infants.

Extubation Readiness in Preterm Infants: Evaluating the Role of Monitoring Intermittent Hypoxemia

Preterm infants with respiratory distress may require mechanical ventilation which is associated with increased pulmonary morbidities. Prompt and successful extubation to noninvasive support is a pressing goal. In this communication, we show original data that increased recurring intermittent hypoxemia (IH, oxygen saturation <80%) may be associated with extubation failure at 72 h in a cohort of neonates <30 weeks gestational age. Current-generation bedside high-resolution pulse oximeters provide saturation profiles that may be of use in identifying extubation readiness and failure. A larger prospective study that utilizes intermittent hypoxemia as an adjunct predictor for extubation readiness is warranted.

Diaphragmatic activity and neural breathing variability during a 5-min endotracheal continuous positive airway pressure trial in extremely preterm infants

BACKGROUND

Extremely preterm infants are often exposed to endotracheal tube continuous positive airway pressure (ETT-CPAP) trials to assess extubation readiness. The effects of ETT-CPAP trial on their diaphragmatic activity (Edi) and breathing variability is unknown.

METHODS

Prospective observational study enrolling infants with birth weight ≤1250 g undergoing their first extubation attempt. Diaphragmatic activity, expressed as the absolute minimum (Edi min) and maximum values (Edi max), area under the Edi signal, and breath-by-breath analyses for breath areas, amplitudes, widths, and neural inspiratory and expiratory times, were analyzed during mechanical ventilation (MV) and ETT-CPAP. Neural breathing variability of each of these parameters was also calculated and compared between MV and ETT-CPAP.

RESULTS

Thirteen infants with median (interquartile range) birth weight of 800 g [610-920] and gestational age of 25.4 weeks [24.4-26.3] were included. Diaphragmatic activity significantly increased during ETT-CPAP when compared to MV:Edi max (44.2 vs. 38.1 μV), breath area (449 vs. 312 μV·s), and amplitude (10.12 vs. 7.46 μV). Neural breathing variability during ETT-CPAP was characterized by increased variability for amplitude and area under the breath, and decreased for breath time and width.

CONCLUSIONS

A 5-min ETT-CPAP in extremely preterm infants undergoing extubation imposed significant respiratory load with changes in respiratory variability.

IMPACT

ETT-CPAP trials are often used to assess extubation readiness in extremely preterm infants, but its effects upon their respiratory system are not well known. Diaphragmatic activity analysis demonstrated that these infants are able to mount an important response to a short trial. A 5-min trial imposed a significant respiratory load evidenced by increased diaphragmatic activity and changes in breathing variability. Differences in breathing variability were observed between successful and failed extubations, which should be explored further in extubation readiness investigations. This type of trial cannot be recommended for preterm infants in clinical practice until clear standards and accuracy are established.

Spontaneous Breathing Trials in Preterm Infants: Systematic Review and Meta-Analysis

BACKGROUND

Few data are available on the use of spontaneous breathing trials (SBTs) in the neonatal population, despite advocacy of the practice in many neonatal ICUs. In this meta-analysis, we systematically reviewed the literature regarding the accuracy of SBTs as a predictor for extubation failure in premature infants.

METHODS

Following the PRISMA recommendations, scientific articles were collected in December 2019 and January 2020 using PubMed, LILACS, Web of Science, Scopus, Google Scholar, OATD, and BDTD databases. The risk of bias in the studies included herein was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 tool. The pooled sensitivity and specificity of the studies were estimated using a mixed logistic regression model of 2 levels and a normal bivariate model.

RESULTS

Six studies were included for qualitative and quantitative evaluation in this study. All SBTs were performed with endotracheal CPAP, with a total observation time of 3-5 min. The parameters for passing/failing the test were similar in 5 of the 6 studies and included bradycardia or desaturation during the test. The SBT showed a high pooled sensitivity (0.97, 95% CI 0.85-0.99), indicating proper identification of neonates "ready" for successful extubation. However, a low pooled specificity (0.40, 95% CI 0.24-0.58), with many false-positive cases, indicated inaccurate prediction of extubation failure. Heterogeneity of included studies was considerable for sensitivity and substantial for specificity.

CONCLUSIONS

The SBT in premature infants can accurately predict extubation success but not extubation failure. Therefore, even though it is an attractive, practical, and easy-to-perform bedside assessment tool, there is a lack of evidence to support its use as an independent predictor of extubation failure in premature infants. Its routine use should be evaluated and monitored carefully.

Cardiorespiratory behavior of preterm infants receiving continuous positive airway pressure and high flow nasal cannula post extubation: randomized crossover study

BACKGROUND

Nasal continuous positive airway pressure (NCPAP) and high flow nasal cannula (HFNC) are modes of non-invasive respiratory support commonly used after extubation in extremely preterm infants. However, the cardiorespiratory physiology of these infants on each mode is unknown.

METHODS

Prospective, randomized crossover study in infants with birth weight ≤1250 g undergoing their first extubation attempt. NCPAP and HFNC were applied randomly for 45 min each, while ribcage and abdominal movements, electrocardiogram, oxygen saturation, and fraction of inspired oxygen (FiO₂) were recorded. Respiratory signals were analyzed using an automated method, and differences between NCPAP and HFNC features and changes in FiO₂ were analyzed.

RESULTS

A total of 30 infants with median [interquartile range] gestational age of 27 weeks [25.7, 27.9] and birth weight of 930 g [780, 1090] were studied. Infants were extubated at 5 days [2, 13] of life with 973 g [880, 1170] and three failed (10%). No differences in cardiorespiratory behavior were noted, except for longer respiratory pauses (9.2 s [5.0, 11.5] vs. 7.3 s [4.6, 9.3]; p = 0.04) and higher FiO₂ levels (p = 0.02) during HFNC compared to NCPAP.

CONCLUSIONS

In extremely preterm infants studied shortly after extubation, the use of HFNC was associated with longer respiratory pauses and higher FiO₂ requirements.

Expired Tidal Volume Variation in Extremely Low Birth Weight and Very Low Birth Weight Infants on Volume-Targeted Ventilation

In a prospective study we describe the delivery of small tidal volumes to extremely low birth weight (ELBW) and very low birth weight (VLBW) infants using a volume-targeted ventilation mode (VTV). Tidal volume delivery was consistent for both ELBW and VLBW infants independent of gestational age, birth weight, and the target volume.

Predictors of extubation readiness in preterm infants: a systematic review and meta-analysis

CONTEXT

A variety of extubation readiness tests have already been incorporated into clinical practice in preterm infants.

OBJECTIVE

To identify predictor tests of successful extubation and determine their accuracy compared with clinical judgement alone.

METHODS

MEDLINE, Embase, PubMed, Cochrane Library and Web of Science were searched between 1984 and June 2016. Studies evaluating predictors of extubation success during a period free of mechanical inflations in infants less than 37 weeks' gestation were included. Risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 tool. After identifying and describing all predictor tests, pooled sensitivity and specificity estimates for the different test categories were generated using a bivariate random-effects model.

RESULTS

Thirty-five studies were included, showing wide heterogeneities in population characteristics, methodologies and definitions of extubation success. Assessments ranged from a few seconds to 24 hours, provided 0-6 cmH₂O positive end-expiratory pressure and measured several clinical and/or physiological parameters. Thirty-one predictor tests were identified, showing good sensitivities but low and variable specificities. Given the high variation in test definitions across studies, pooling could only be performed on a subset. The commonly performed spontaneous breathing trials had pooled sensitivity of 95% (95% CI 87% to 99%) and specificity of 62% (95% CI 38% to 82%), while composite tests offered the best performance characteristics.

CONCLUSIONS

There is a lack of strong evidence to support the use of extubation readiness tests in preterm infants. Although spontaneous breathing trials are attractive assessment tools, higher quality studies are needed for determining the optimal strategies for improving their accuracy.

Prediction of Extubation readiness in extremely preterm infants by the automated analysis of cardiorespiratory behavior: study protocol

Background

Extremely preterm infants (≤ 28 weeks gestation) commonly require endotracheal intubation and mechanical ventilation (MV) to maintain adequate oxygenation and gas exchange. Given that MV is independently associated with important adverse outcomes, efforts should be made to limit its duration. However, current methods for determining extubation readiness are inaccurate and a significant number of infants fail extubation and require reintubation, an intervention that may be associated with increased morbidities. A variety of objective measures have been proposed to better define the optimal time for extubation, but none have proven clinically useful. In a pilot study, investigators from this group have shown promising results from sophisticated, automated analyses of cardiorespiratory signals as a predictor of extubation readiness. The aim of this study is to develop an automated predictor of extubation readiness using a combination of clinical tools along with novel and automated measures of cardiorespiratory behavior, to assist clinicians in determining when extremely preterm infants are ready for extubation.

Methods

In this prospective, multicenter observational study, cardiorespiratory signals will be recorded from 250 eligible extremely preterm infants with birth weights ≤1250 g immediately prior to their first planned extubation. Automated signal analysis algorithms will compute a variety of metrics for each infant, and machine learning methods will then be used to find the optimal combination of these metrics together with clinical variables that provide the best overall prediction of extubation readiness. Using these results, investigators will develop an Automated system for Prediction of EXtubation (APEX) readiness that will integrate the software for data acquisition, signal analysis, and outcome prediction into a single application suitable for use by medical personnel in the neonatal intensive care unit. The performance of APEX will later be prospectively validated in 50 additional infants.

Discussion

The results of this research will provide the quantitative evidence needed to assist clinicians in determining when to extubate a preterm infant with the highest probability of success, and could produce significant improvements in extubation outcomes in this population.

Trial registration

Clinicaltrials.gov identifier: NCT01909947. Registered on July 17 2013.

Trial sponsor: Canadian Institutes of Health Research (CIHR).

Electronic supplementary material

The online version of this article (doi:10.1186/s12887-017-0911-z) contains supplementary material, which is available to authorized users.

APEX_SCOPE: A graphical user interface for visualization of multi-modal data in inter-disciplinary studies

In multi-disciplinary studies, different forms of data are often collected for analysis. For example, APEX, a study on the automated prediction of extubation readiness in extremely preterm infants, collects clinical parameters and cardiorespiratory signals. A variety of cardiorespiratory metrics are computed from these signals and used to assign a cardiorespiratory pattern at each time. In such a situation, exploratory analysis requires a visualization tool capable of displaying these different types of acquired and computed signals in an integrated environment. Thus, we developed APEX_SCOPE, a graphical tool for the visualization of multi-modal data comprising cardiorespiratory signals, automated cardiorespiratory metrics, automated respiratory patterns, manually classified respiratory patterns, and manual annotations by clinicians during data acquisition. This MATLAB-based application provides a means for collaborators to view combinations of signals to promote discussion, generate hypotheses and develop features.

Basic principles of respiratory function monitoring in ventilated newborns: A review

Respiratory monitoring during mechanical ventilation provides a real-time picture of patient-ventilator interaction and is a prerequisite for lung-protective ventilation. Nowadays, measurements of airflow, tidal volume and applied pressures are standard in neonatal ventilators. The measurement of lung volume during mechanical ventilation by tracer gas washout techniques is still under development. The clinical use of capnography, although well established in adults, has not been embraced by neonatologists because of technical and methodological problems in very small infants. While the ventilatory parameters are well defined, the calculation of other physiological parameters are based upon specific assumptions which are difficult to verify. Incomplete knowledge of the theoretical background of these calculations and their limitations can lead to incorrect interpretations with clinical consequences. Therefore, the aim of this review was to describe the basic principles and the underlying assumptions of currently used methods for respiratory function monitoring in ventilated newborns and to highlight methodological limitations.

Pediatric Pulmonology year in review 2015: Part 1

Our journal covers a broad range of research and scholarly topics related to children's respiratory disorders. For updated perspectives on the rapidly expanding knowledge in our field, we will summarize the past year's publications in our major topic areas, as well as selected publications in these areas from the core clinical journal literature outside our own pages. The current review covers articles on neonatal lung disease, pulmonary physiology, and respiratory infection. Pediatr Pulmonol. 2016;51:733-739. © 2016 Wiley Periodicals, Inc.

Correlation of clinical parameters with cardiorespiratory behavior in successfully extubated extremely preterm infants

Extremely preterm infants (gestational age ≤ 28 weeks) often require EndoTracheal Tube-Invasive Mechanical Ventilation (ETT-IMV) to survive. Clinicians wean infants off ETT-IMV as early as possible using their judgment and clinical information. However, assessment of extubation readiness is not accurate since 20 to 40% of preterm infants fail extubation. We extended our work in automated prediction of extubation readiness by examining correlations of automated cardiorespiratory features to clinical parameters in successfully extubated infants. Only a few features, mainly those related to variability of breathing synchrony, had any consistent correlation with clinical parameters, namely gestational age, day of life at extubation, and bicarbonate. We conclude that the automated cardiorespiratory features likely provide different information additional to clinical practice.

Scoring Tools for the Analysis of Infant Respiratory Inductive Plethysmography Signals

Infants recovering from anesthesia are at risk of life threatening Postoperative Apnea (POA). POA events are rare, and so the study of POA requires the analysis of long cardiorespiratory records. Manual scoring is the preferred method of analysis for these data, but it is limited by low intra- and inter-scorer repeatability. Furthermore, recommended scoring rules do not provide a comprehensive description of the respiratory patterns. This work describes a set of manual scoring tools that address these limitations. These tools include: (i) a set of definitions and scoring rules for 6 mutually exclusive, unique patterns that fully characterize infant respiratory inductive plethysmography (RIP) signals; (ii) RIPScore, a graphical, manual scoring software to apply these rules to infant data; (iii) a library of data segments representing each of the 6 patterns; (iv) a fully automated, interactive formal training protocol to standardize the analysis and establish intra- and inter-scorer repeatability; and (v) a quality control method to monitor scorer ongoing performance over time. To evaluate these tools, three scorers from varied backgrounds were recruited and trained to reach a performance level similar to that of an expert. These scorers used RIPScore to analyze data from infants at risk of POA in two separate, independent instances. Scorers performed with high accuracy and consistency, analyzed data efficiently, had very good intra- and inter-scorer repeatability, and exhibited only minor confusion between patterns. These results indicate that our tools represent an excellent method for the analysis of respiratory patterns in long data records. Although the tools were developed for the study of POA, their use extends to any study of respiratory patterns using RIP (e.g., sleep apnea, extubation readiness). Moreover, by establishing and monitoring scorer repeatability, our tools enable the analysis of large data sets by multiple scorers, which is essential for longitudinal and multicenter studies.