Predicting apneic events in preterm infants using cardio-respiratory and movement features

Abstract

Background and objective: Preterm neonates are prone to episodes of apnea, bradycardia and hypoxia (ABH) that can lead to neurological morbidities or even death. There is broad interest in developing methods for real-time prediction of ABH events to inform interventions that prevent or reduce their incidence and severity. Using advances in machine learning methods, this study develops an algorithm to predict ABH events.

Methods: Following previous studies showing that respiratory instabilities are closely associated with bouts of movement, we present a modeling framework that can predict ABH events using both movement and cardio-respiratory features derived from routine clinical recordings. In 10 preterm infants, movement onsets and durations were estimated with a wavelet-based algorithm that quantified artifactual distortions of the photoplethysmogram signal. For prediction, cardio-respiratory features were created from time-delayed correlations of inter-beat and inter-breath intervals with past values; movement features were derived from time-delayed correlations with inter-breath intervals. Gaussian Mixture Models and Logistic Regression were used to develop predictive models of apneic events. Performance of the models was evaluated with ROC curves.

Results: Performance of the prediction framework (mean AUC) was 0.77 ± 0.04 for 66 ABH events on training data from 7 infants. When grouped by the severity of the associated bradycardia during the ABH event, the framework was able to predict 83% and 75% of the most severe episodes in the 7-infant training set and 3-infant test set, respectively. Notably, inclusion of movement features significantly improved the predictions compared with modeling with only cardio-respiratory signals.

Conclusions: Our findings suggest that recordings of movement provide important information for predicting ABH events in preterm infants, and can inform preemptive interventions designed to reduce the incidence and severity of ABH events.

Fig. 1.

Raw signals of respiratory activity, ECG, SPO₂ and PPG. Binary markers of movements were derived from the PPG signal. Apneas and the associated bradycardias and oxygen desaturation are highlighted.

Fig. 2.

Signals used for feature engineering. Left: Time-series of RR intervals, inter-breath intervals (IBI) and movement. Right: Log-transformed and interpolated versions of RR intervals and IBIs.

Fig. 3.

Labeling of apnea, pre-apnea and inter-apnea blocks.

Fig. 4.

Schematic representation for feature engineering.

Fig. 5.

Example of stratified k-fold cross-validation.

Fig. 6.

Schematic Diagram of the classifier based on Gaussian Mixture Models (GMM) and Logistic Regression.

Fig. 7.

Real-time prediction scheme.

Fig. 8.

ROC curves from prediction scores for Subjects 1 and 2. The dashed red line signifies random classification. The shaded green area represents the standard deviation of the ROC curve obtained from 50 repeated runs of the model.

Fig. 9.

Probability distribution of the AUC from 1000 shuffled runs for Subjects 1 and 2. The original AUC is indicated by the dashed red lines.

Fig. 10.

Cumulative distribution of prediction time relative to the onset of apnea (training data set).

Fig. 11.

Cross-validated prediction based on severity of apneas (training data set).

Fig. 12.

Comparison of performance methods from models including (blue) and excluding (red) movement-derived features. The black dashed line in the AUC Fig. indicates classification by chance. When movement features were excluded to train the models, classification within the subjects #1, 4, 5, 6 and 7 were not significantly different from chance. All other models were significant.

Fig. 13.

Cumulative distribution of prediction time relative to onset of apnea (test data set).

Fig. 14.

Prediction based on severity of apneas (test data set).