Automated Oxygen Delivery in Neonatal Intensive Care

Oxygenation Fluctuations Associated with Severe Retinopathy of Prematurity: Insights from a Multimodal Deep Learning Approach

Purpose

Retinopathy of prematurity (ROP) is one of the leading causes of blindness in children. Although the role of oxygen in the pathophysiology of ROP is well established, a precise understanding of the dynamic relationship between oxygen exposure ROP incidence and severity is lacking. The purpose of this study was to evaluate the correlation between time-dependent oxygen variables and the onset of ROP.

Design

Retrospective cohort study.

Participants

Two hundred thirty infants who were born at a single academic center and met the inclusion criteria were included. Infants are mainly born between January 2011 and October 2022.

Methods

Patient data were extracted from electronic health records (EHRs), with sufficient time-dependent oxygen data. Clinical outcomes for ROP were recorded as none/mild or moderate/severe (defined as type II or worse). Mixed-effects linear models were used to compare the 2 groups in terms of dynamic oxygen variables, such as daily average and the coefficient of variation (COV) fraction of inspired oxygen (FiO2). Support vector machine (SVM) and long-short-term memory (LSTM)-based multimodal models were trained with fivefold cross-validation to predict which infants would develop moderate/severe ROP. Gestational age (GA), birth weight, and time-dependent oxygen variables were used to develop predictive models.

Main Outcome Measures

Model cross-validation performance was evaluated by computing the mean area under the receiver operating characteristic (AUROC) curve, precision, recall, and F1 score.

Results

We found that both daily average and COV of FiO2 were associated with more severe ROP (adjusted P < 0.001). With fivefold cross-validation, the multimodal LSTM models had higher performance than the best static models (SVM using GA and 3 average FiO2 features) and SVM models trained on GA alone (mean AUROC = 0.89 ± 0.04 vs. 0.86 ± 0.05 vs. 0.83 ± 0.04).

Conclusions

The development of severe ROP might not only be influenced by oxygen exposure but also by its fluctuation, which provides direction for future study of pathophysiological factors associated with severe ROP development. Additionally, we demonstrated that multimodal neural networks can be a method to extract useful information from time-series data, which may be a valuable methodology for the investigation of other diseases using EHR data.

Financial Disclosures

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Monitoring SpO2: The Basics of Retinopathy of Prematurity (Back to Basics) and Targeting Oxygen Saturation

Oxygen (O2) is a drug frequently used in newborn care. Adverse effects of hypoxia are well known but the damaging effects of excess oxygen administration and oxidative stress have only been studied in the last 2 decades. Many negative effects have been described, including retinopathy of prematurity . Noninvasive pulse oximetry (SpO2) is useful to detect hypoxemia but requires careful evaluation and understanding of the frequently changing relationship between O2 and hemoglobin to prevent hyperoxemia. Intention to treat SpO2 ranges should be individualized for every newborn receiving supplemental O2, according to gestational age, post-natal age, and clinical condition.

Prolonged use of closed-loop inspired oxygen support in preterm infants: a randomised controlled trial

OBJECTIVE

This randomised study in preterm infants on non-invasive respiratory support investigated the effectiveness of automated oxygen control (A-FiO2) in keeping the oxygen saturation (SpO2) within a target range (TR) during a 28-day period compared with manual titration (M-FiO2).

DESIGN

A single-centre randomised control trial.

SETTING

A level III neonatal intensive care unit.

PATIENTS

Preterm infants (<28 weeks' gestation) on non-invasive respiratory support.

INTERVENTIONS

A-FiO2 versus M-FiO2 control.

METHODS

Main outcomes were the proportion of time spent and median area of episodes in the TR, hyperoxaemia, hypoxaemia and the trend over 28 days using a linear random intercept model.

RESULTS

23 preterm infants (median gestation 25.7 weeks; birth weight 820 g) were randomised. Compared with M-FiO2, the time spent within TR was higher in the A-FiO2 group (68.7% vs 48.0%, p<0.001). Infants in the A-FiO2 group spent less time in hyperoxaemia (13.8% vs 37.7%, p<0.001), but no difference was found in hypoxaemia. The time-based analyses showed that the A-FiO2 efficacy may differ over time, especially for hypoxaemia. Compared with the M-FiO2 group, the A-FiO2 group had a larger intercept but with an inversed slope for the daily median area below the TR (intercept 70.1 vs 36.3; estimate/day -0.70 vs 0.69, p<0.001).

CONCLUSION

A-FiO2 control was superior to manual control in keeping preterm infants on non-invasive respiratory support in a prespecified TR over a period of 28 days. This improvement may come at the expense of increased time below the TR in the first days after initiating A-FiO2 control.

TRIAL REGISTRATION NUMBER

NTR6731.

Randomised control trial of oxygen assist module in preterm infants on high-flow nasal cannula support

OBJECTIVE

To evaluate the efficacy of automatic oxygen control (A-FiO2) in reducing the extremes of oxygen saturations (SpO2<80% and SpO2>98%) in preterm infants on high-flow nasal cannula (HFNC) respiratory support using Vapotherm Precision Flow.

DESIGN

A parallel-arm randomised controlled trial.

SETTING

A level-III neonatal intensive care unit.

PATIENTS

Preterm infants born <33 (23+0 to 32+6) weeks receiving HFNC as respiratory support.

INTERVENTIONS

A-FiO2 versus manual (M-FiO2) oxygen control during the full course of HFNC support.

OUTCOMES

The primary outcome of this study is percentage of time spent in extreme oxygen saturations (<80% and >98%) in preterm infants when receiving HFNC as respiratory support. Secondary outcomes were time with SpO2 between 90% and 95% plus time >95% without supplemental oxygen.

RESULTS

60 infants were randomised equally to either A-FiO2 or M-FiO2 arm. Their baseline characteristics were comparable. They spent a median of 5.3 (IQR: 2.0-8.4) and 6.5 (IQR: 2.9-13.7) days in the study, A-FiO2 and M-FiO2, respectively. The percentage of time spent in SpO2<80% (median of 0.4% (0.1%-0.8%) vs 1.6% (0.6%-2.6%), p=0.002) and >98% (median 0.2% (0.1%-0.9%) vs 1.9% (0.7%-4%), p<0.001) were significantly lower in A-FiO2 compared with M-FiO2. The difference in median percentage of time in target range between the two arms was 26% (81% (74%-93%) in A-FiO2 vs 55% (48%-72%) in M-FiO2).

CONCLUSION

A-FiO2 was associated with statistically significant reduction in the percentage of time spent in extremes of saturation when compared with M-FiO2 in preterm infants receiving HFNC.

TRIAL REGISTRATION NUMBER

NCT04687618.

The critical role of technologies in neonatal care

Neonatal care has made significant advances in the last few decades. As a result, mortality and morbidity in high-risk infants, such as extremely preterm infants or those infants with birth-related brain injury, has reduced significantly. Many of these advances have been facilitated or delivered through development of medical technologies allowing clinical teams to be better supported with the care they deliver or provide new therapies and diagnostics to improve management. The delivery of neonatal intensive care requires the provision of medical technologies that are easy to use, reliable, accurate and ideally developed for the unique needs of the newborn population. Many technologies have been developed and commercialised following adult trials without ever being studied in neonatal patients despite the unique characteristics of this population. Increasingly, funders and industry are recognising this major challenge which has resulted in initiatives to develop new ideas from concept through to clinical care. This review explores some of the key medical technologies used in neonatal care and the evidence to support their adoption to improve outcomes. A number of devices have yet to realise their full potential and will require further development to optimise and find their ideal target population and clinical benefit. Examples of emerging technologies, which may soon become more widely used, are also discussed. As neonatal care relies more on medical technologies, we need to be aware of the impact on care pathways, especially from a human factors approach, the associated costs and subsequent benefits to patients alongside the supporting evidence.

Determining an Optimal Oxygen Saturation Target Range Based on Neonatal Maturity: Demonstration of a Decision Tree Analytic

The utility of decision tree machine learning in exploring the interactions among the SpO2 target range, neonatal maturity, and oxemic-risk is demonstrated.

METHODS

This observational study used 3 years of paired age-SpO2-PaO2 data from a neonatal ICU. The CHAID decision tree method was used to explore the interaction of postmenstrual age (PMA) on the risk of extreme arterial oxygen levels at six different potential SpO2 target ranges (88-92%, 89-93%, 90-94%, 91-95%, 92-96% and 93-97%). Risk was calculated using a severity-weighted average of arterial oxygen outside the normal range for neonates (50-80 mmHg).

RESULTS

In total, 7500 paired data points within the potential target range envelope were analyzed. The two lowest target ranges were associated with the highest risk, and the ranges of 91-95% and 92-96% were associated with the lowest risk. There were shifts in the risk associated with PMA. All the target ranges showed the lowest risk at ≥42 weeks PMA. The lowest risk for preterm infants was within a target range of 92-96% with a PMA of ≤34 weeks.

CONCLUSIONS

This study demonstrates the utility of decision tree analytics. These results suggest that SpO2 target ranges that are different from typical range might reduce morbidity and mortality. Further research, including prospective randomized trials, is warranted.

Routine use of automated FiO2 control in Poland: prospective registry and survey

Objective

The performance of automated control of inspired oxygen (A-FiO2) has been confirmed in dozens of studies but reports of routine use are limited. Broadly adopted in Poland, our aim is to share that experience.

Methods

We used a prospectively planned observational study of the performance, general use patterns, unit practices, and problems with A-FiO2, based on a web registry of case reports, complemented by surveys of subjective impressions.

Results

In 2019, a total of 92 A-FiO2 systems were in routine use in 38 centers. Of the 38 centers, 20 had agreed in 2013 to participate in the project. In these centers, A-FiO2 was applied in infants of all weights, but some centers restricted its use to weaning from oxygen and unstable infants. A cohort had reported their experience with each use (5/20 centers, 593 cases). A quarter of those infants were managed with a lower target range and three-quarters with alarms looser than European guidelines for manual SpO2 control. The perceived primary advantages of A-FiO2 were as follows: keeping the readings in the target range, reducing exposure to SpO2 extremes, reducing risk from nurse distraction, reducing workload, and reducing alarm fatigue. Practices did evolve with experience, including implementing changes in the alarm strategy, indications for use, and target range. The potential for over-reliance on automation was cited as a risk. There were a few reports of limited effectiveness (moderate 12/593 and poor 2/593).

Conclusions

Automated oxygen control is broadly perceived by users as an improvement in controlling SpO2 with infrequent problems.

Closed-loop oxygen system in late preterm/term, ventilated infants with different severities of respiratory disease

AIM

To evaluate closed-loop automated oxygen control (CLAC) in ventilated infants >33 weeks of gestation with different respiratory disease severities.

METHODS

Infants were studied on two consecutive days for 6 h each day. They were randomised to receive standard care or standard care with CLAC (Oxygenie) first. Analyses were performed of the results of infants with or without an FiO₂  ≥ 0.3 and infants with congenital diaphragmatic hernia (CDH).

RESULTS

Thirty-one infants with a median (IQR) gestational age of 37.9 (37.1-38.9) weeks were studied at a median postmenstrual age (IQR) of 38.9 (37.4-39.8) weeks. In infants with an FiO₂  ≥ 0.3 (n = 8), CLAC increased the time spent in target oxygen range (92-96%) by 61.6% (p = 0.018), whereas in infants with an FiO₂  < 0.3, the time in target was increased by 3.8% (p = 0.019). During CLAC, only infants with an FiO₂  ≥ 0.3 spent less time in hyperoxemia (SpO₂  > 96%) (p = 0.012) and hyperoxemic episodes were shorter (p = 0.012). In both groups, CLAC reduced the duration of desaturations (SpO₂  < 92%, p < 0.001). In CDH infants, CLAC increased the time spent in target oxygen range by 34% (p = 0.036) and the median duration of desaturations was reduced (p = 0.028).

CONCLUSION

CLAC may be more useful in infants with more severe respiratory distress.

Oxygenation in the NICU: there is more to it than meets the eye

Dormishian and colleagues in their study address an issue that care teams in the NICU encounter on a daily basis, regarding motion artifacts during oxygenation monitoring. In our commentary, we discuss the available tools that allow continuous noninvasive monitoring of oxygenation in the NICU, and modalities that increase the time premature infants spend in the desired SpO₂ range and impact their clinical outcomes.