Detecting central apneas using multichannel signals in premature infants.

Objective. Monitoring of apnea of prematurity, performed in neonatal intensive care units by detecting central apneas (CAs) in the respiratory traces, is characterized by a high number of false alarms. A two-step approach consisting of a threshold-based apneic event detection algorithm followed by a machine learning model was recently presented in literature aiming to improve CA detection. However, since this is characterized by high complexity and low precision, we developed a new direct approach that only consists of a detection model based on machine learning directly working with multichannel signals.Approach. The dataset used in this study consisted of 48 h of ECG, chest impedance and peripheral oxygen saturation extracted from 10 premature infants. CAs were labeled by two clinical experts. 47 features were extracted from time series using 30 s moving windows with an overlap of 5 s and evaluated in sets of 4 consecutive moving windows, in a similar way to what was indicated for the two-step approach. An undersampling method was used to reduce imbalance in the training set while aiming at increasing precision. A detection model using logistic regression with elastic net penalty and leave-one-patient-out cross-validation was then tested on the full dataset.Main results. This detection model returned a mean area under the receiver operating characteristic curve value equal to 0.86 and, after the selection of a FPR equal to 0.1 and the use of smoothing, an increased precision (0.50 versus 0.42) at the expense of a decrease in recall (0.70 versus 0.78) compared to the two-step approach around suspected apneic events.Significance. The new direct approach guaranteed correct detections for more than 81% of CAs with lengthL≥ 20 s, which are considered among the most threatening apneic events for premature infants. These results require additional verifications using more extensive datasets but could lead to promising applications in clinical practice.

A Continuous Late-Onset Sepsis Prediction Algorithm for Preterm Infants Using Multi-Channel Physiological Signals From a Patient Monitor.

The aim of this study is to develop an explainable late-onset sepsis (LOS) prediction algorithm using continuous multi-channel physiological signals that can be applied to a patient monitor for preterm infants in a neonatal intensive care unit (NICU). The algorithm uses features on heart rate variability (HRV), respiration, and motion, based on electrocardiogram (ECG) and chest impedance (CI). In this study, 127 preterm infants were included, of whom 59 were bloodculture-proven LOS patients and 68 were control patients. Features in 24 hours before the onset of sepsis (LOS group), and an age-matched onset time point (control group) were extracted and fed into machine learning classifiers with gestational age and birth weight. We compared the prediction performance of several well-known classifiers using features from different signal channels (HRV, respiration, and motion) individually as well as their combinations. The prediction performance was evaluated using the area under the receiver-operating-characteristics curve (AUC). The best performance was achieved by an extreme gradient boosting classifier combining features from all signal channels, with an AUC of 0.88, a positive predictive value of 0.80, and a negative predictive value of 0.83 during the 6 hours preceding LOS onset. This feasibility study demonstrates the complementary predictive value of motion information in addition to cardiorespiratory information for LOS prediction. Furthermore, visualization of how each feature in the individual patient impacts the algorithm decision strengthen its interpretability. In clinical practice, it is important to motivate clinical interventions and this visualization method can help to support the clinical decision.

Central apnea detection in premature infants using machine learning.

BACKGROUND AND OBJECTIVE: Apnea of prematurity is one of the most common diagnosis in neonatal intensive care units. Apneas can be classified as central, obstructive or mixed. According to the current international standards, minimal fluctuations or absence of fluctuations in the chest impedance (CI) suggest a central apnea (CA). However, automatic detection of reduced CI fluctuations leads to a high number of central apnea-suspected events (CASEs), the majority being false alarms. We aim to improve automatic detection of CAs by using machine learning to optimize detection of CAs among CASEs. METHODS: Using an optimized algorithm for automated detection, all CASEs were detected in a population of 10 premature infants developing late-onset sepsis and 10 age-matched control patients. CASEs were inspected by two clinical experts and annotated as CAs or rejections in two rounds of annotations. A total of 47 features were extracted from the ECG, CI and oxygen saturation signals considering four 30 s-long moving windows, from 30 s before to 15 s after the onset of each CASE, using a moving step size of 5 s. Consecutively, new CA detection models were developed based on logistic regression with elastic net penalty, random forest and support vector machines. Performance was evaluated using both leave-one-patient-out and 10-fold cross-validation considering the mean area under the receiver-operating-characteristic curve (AUROC). RESULTS: The CA detection model based on logistic regression with elastic net penalty returned the highest mean AUROC when features extracted from all four time windows were included, both using leave-one-patient-out and 10-fold cross-validation (mean AUROC of 0.88 and 0.90, respectively). Feature relevance was found to be the highest for features derived from the CI. A threshold for the false positive rate in the mean receiver-operating-characteristic curve equal to 0.3 led to a high percentage of correct detections for all CAs (78.2%) and even higher for CAs followed by a bradycardia (93.4%) and CAs followed by both a bradycardia and a desaturation (95.2%), which are more critical for the well-being of premature infants. CONCLUSIONS: Models based on machine learning can lead to improved CA detection with fewer false alarms.

Effect of Music Therapy Interventions on Heart Rate Variability in Premature Infants.

Premature infants are at risk of developing serious complications after birth. Communicative interventions performed in neonatal intensive care units (NICUs), such as music therapy interventions, can reduce the stress experienced by these infants and promote the development of their autonomic nervous system. In this study we investigated the effects of music therapy interventions, consisting of singing, humming, talking or rhythmic reading, on premature infants by investigating the effects on their heart rate variability (HRV). A total of 27 communicative intervention from 18 patients were included in this study. The NN-intervals were extracted from the ECG and the mean ± SEM values for the 6 different features (HR, SDNN, RMSSD, pNN50, pDec and SDDec) was investigated. Median feature values for the pre- and communicative intervention were compared using the Wilcoxon signed-rank test. An increase in values for the SDNN, RMSSD and pNN50 was found in the 20 minutes preceding the communicative intervention, when caregiving activities were performed, and was followed by an immediate decrease at the start of the intervention. Features' variability during the intervention appeared to be smaller than in the pre-communicative intervention, indicating improved autonomic regulation. This difference was, however, not statistically significant possibly due to different types of activities applied during the communicative intervention per patient.

The effect of apnea length on vital parameters in apnea of prematurity - Hybrid observations from clinical data and simulation in a mathematical model.

Apnea of prematurity (AOP) is a critical condition for preterm infants which can lead to several adverse outcomes. Despite its relevance, mechanisms underlying AOP are still unclear. In this work we aimed at improving the understanding of AOP and its physiologic responses by analyzing and comparing characteristics of real infant data and model-based simulations of AOP. We implemented an existing algorithm to extract apnea events originating from the central nervous system from a population of 26 premature infants (1248 h of data in total) and investigated oxygen saturation (SpO2) and heart rate (HR) of the infants around these events. We then extended a previously developed cardio-vascular model to include the lung mechanics and gas exchange. After simulating the steady state of a preterm infant, which successfully replicated results described in previous literature studies, the extended model was used to simulate apneas with different lengths caused by a stop in respiratory muscles. Apneas identified by the algorithm and simulated by the model showed several similarities, including a far deeper decrease in SpO2, with the minimum reached later in time, in case of longer apneas. Results also showed some differences, either due to how measures are performed in clinical practice in our neonatal intensive care unit (e.g. delayed detection of decline in SpO2 after apnea onset due to signal averaging) or to the limited number of very long apneas (≥80 s) identified in our dataset.

Optimized Detection of Central Apneas Preceding Late-Onset Sepsis in Premature Infants.

In neonatal intensive care units, respiratory traces of premature infants developing late onset sepsis (LOS) may also show episodes of apneas. However, since clinical patient monitors often underdetect apneas, clinical experts are required to investigate patients' traces looking for these events. In this work we present a method to optimize an existing algorithm for central apnea (CA) detection and how we used it together with human annotations to investigate the occurrence of CAs preceding LOS.The algorithm was optimized by using a previously-annotated dataset consisting of 90 hours, extracted from 10 premature infants. This allowed to double precision (19.7% vs 9.3%, median values per patient) without affecting recall (90.5% vs 94.5%) compared to the original algorithm. This choice caused the missed identification of just 1 additional CA (4 vs 3) in the whole dataset. The optimized algorithm was then used to annotate a second dataset consisting of 480 hours, extracted from 10 premature infants diagnosed with LOS. Annotations were corrected by two clinical experts.A significantly higher number of CA annotations was found in the 6 hours prior to sepsis onset (p-value < 0.05). The use of the optimized algorithm followed by human annotations proved to be a suitable, time-efficient method to annotate CAs before sepsis in premature infants, enabling future use in large datasets.

Optimisation of clinical workflow and monitor settings safely reduces alarms in the NICU.

AIM: To address alarm fatigue, a new alarm management system which ensures a quicker delivery of alarms together with waveform information on nurses' handheld devices was implemented and settings optimised. The effects of this clinical implementation on alarm rates and nurses' responsiveness were measured in an 18-bed single family rooms neonatal intensive care unit (NICU). METHODS: The technical implementation of the alarm management system was followed by clinical workflow optimisation. Alarms and vital parameters from October 2017 to December 2019 were analysed. Measures included monitoring alarms, nurses' response to alarms and time spent by patients in different saturation ranges. A survey among nurses was performed to evaluate changes in alarm rate and use of protocols. RESULTS: A significant reduction of monitoring alarms per patient days was detected after the optimisation phase (in particular for SpO2 ≤ 80%, P < .001). More time was spent by infants within the optimal peripheral oxygen saturation range (88% < SpO2 < 95%, P < .001). Results from the surveys showed that false alarms are less likely to cause an inappropriate response after the optimisation phase. CONCLUSION: The implementation of an alarm management solution and an optimisation programme can safely reduce the alarm burden inside of the NICU environment.

Estimation of Heart Rate Directly from ECG Spectrogram in Neonate Intensive Care Units.

This paper presents a simple yet novel method to estimate the heart frequency (HF) of neonates directly from the ECG signal, instead of using the RR-interval signals as generally done in clinical practices. From this, the heart rate (HR) can be derived. Thus, we avoid the use of peak detectors and the inherent errors that come with them.Our method leverages the highest Power Spectral Densities (PSD) of the ECG, for the bins around the frequencies related to heart rates for neonates, as they change in time (spectrograms).We tested our approach with the monitoring data of 6 days for 52 patients in a Neonate Intensive Care Unit (NICU) and compared against the HR from a commercial monitor, which produced a sample every second. The comparison showed that 92.4% of the samples have a difference lower than 5bpm. Moreover, we obtained a median MAE (Mean Absolute Error) between subjects equal to 2.28 bpm and a median RMSE (Root Mean Square Error) equal to 5.82 bpm. Although tested for neonates, we hypothesize that this method can also be customized for other populations.Finally, we analyze the failure cases of our method and found a direct co-allocation of errors due to moments with higher PSD in the lower frequencies with the presence of critical alarms related to other physiological systems (e.g. desaturation).