A semi-Markov chain approach to modeling respiratory patterns prior to extubation in preterm infants

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.

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.

Multi-Chain Semi-Markov Analysis of Intrapartum Cardiotocography

Visual assessment of the evolution of fetal heart rate (FHR) and uterine pressure (UP) patterns is the standard of care in the intrapartum period. Unfortunately, this assessment has high levels of intra- and inter-observer variability. This study processed and analyzed FHR and UP patterns using computerized pattern recognition tools. The goal was to evaluate differences in FHR and UP patterns between fetuses with normal outcomes and those who developed hypoxic-ischemic encephalopathy (HIE). For this purpose, we modeled the sequence of FHR patterns and uterine contractions using Multi-Chain Semi-Markov models (MCSMMs). These models estimate the probability of transitioning between FHR or UP patterns and the dwell time of each pattern. Our results showed that in comparison to the control group, the HIE group had: (1) more frequent uterine contractions during the last 12 hours before birth; (2) more frequent FHR decelerations during the last 12 hours before birth; (3) longer decelerations during the last eight hours before birth; and (4) shorter baseline durations during the last five hours before birth. These results demonstrate that the fetuses in the HIE group were subject to a more stressful environment than those in the normal group. Clinical Relevance- Our results revealed statistically significant differences in FHR/UP patterns between the normal and HIE groups in the hours before birth. This indicates that features derived using MCSMMs may be useful in a machine learning framework to detect infants at increased risk of developing HIE allowing preventive interventions.

Automatic unsupervised respiratory analysis of infant respiratory inductance plethysmography signals

Infants are at risk for potentially life-threatening postoperative apnea (POA). We developed an Automated Unsupervised Respiratory Event Analysis (AUREA) to classify breathing patterns obtained with dual belt respiratory inductance plethysmography and a reference using Expectation Maximization (EM). This work describes AUREA and evaluates its performance. AUREA computes six metrics and inputs them into a series of four binary k-means classifiers. Breathing patterns were characterized by normalized variance, nonperiodic power, instantaneous frequency and phase. Signals were classified sample by sample into one of 5 patterns: pause (PAU), movement (MVT), synchronous (SYB) and asynchronous (ASB) breathing, and unknown (UNK). MVT and UNK were combined as UNKNOWN. Twenty-one preprocessed records obtained from infants at risk for POA were analyzed. Performance was evaluated with a confusion matrix, overall accuracy, and pattern specific precision, recall, and F-score. Segments of identical patterns were evaluated for fragmentation and pattern matching with the EM reference. PAU exhibited very low normalized variance. MVT had high normalized nonperiodic power and low frequency. SYB and ASB had a median frequency of respectively, 0.76Hz and 0.71Hz, and a mode for phase of 4^o and 100^o. Overall accuracy was 0.80. AUREA confused patterns most often with UNKNOWN (25.5%). The pattern specific F-score was highest for SYB (0.88) and lowest for PAU (0.60). PAU had high precision (0.78) and low recall (0.49). Fragmentation was evident in pattern events <2s. In 75% of the EM pattern events >2s, 50% of the samples classified by AUREA had identical patterns. Frequency and phase for SYB and ASB were consistent with published values for synchronous and asynchronous breathing in infants. The low normalized variance in PAU, was consistent with published scoring rules for pediatric apnea. These findings support the use of AUREA to classify breathing patterns and warrant a future evaluation of clinically relevant respiratory events.