Discovery of signatures of fatal neonatal illness in vital signs using highly comparative time-series analysis

Highly comparative time series analysis of oxygen saturation and heart rate to predict respiratory outcomes in extremely preterm infants

Objective.Highly comparative time series analysis (HCTSA) is a novel approach involving massive feature extraction using publicly available code from many disciplines. The Prematurity-Related Ventilatory Control (Pre-Vent) observational multicenter prospective study collected bedside monitor data from>700extremely preterm infants to identify physiologic features that predict respiratory outcomes.Approach. We calculated a subset of 33 HCTSA features on>7 M 10 min windows of oxygen saturation (SPO2) and heart rate (HR) from the Pre-Vent cohort to quantify predictive performance. This subset included representatives previously identified using unsupervised clustering on>3500HCTSA algorithms. We hypothesized that the best HCTSA algorithms would compare favorably to optimal PreVent physiologic predictor IH90_DPE (duration per event of intermittent hypoxemia events below 90%).Main Results.The top HCTSA features were from a cluster of algorithms associated with the autocorrelation of SPO2 time series and identified low frequency patterns of desaturation as high risk. These features had comparable performance to and were highly correlated with IH90_DPE but perhaps measure the physiologic status of an infant in a more robust way that warrants further investigation. The top HR HCTSA features were symbolic transformation measures that had previously been identified as strong predictors of neonatal mortality. HR metrics were only important predictors at early days of life which was likely due to the larger proportion of infants whose outcome was death by any cause. A simple HCTSA model using 3 top features outperformed IH90_DPE at day of life 7 (.778 versus .729) but was essentially equivalent at day of life 28 (.849 versus .850).Significance. These results validated the utility of a representative HCTSA approach but also provides additional evidence supporting IH90_DPE as an optimal predictor of respiratory outcomes.

Highly comparative time series analysis of oxygen saturation and heart rate to predict respiratory outcomes in extremely preterm infants

Objective

Highly comparative time series analysis (HCTSA) is a novel approach involving massive feature extraction using publicly available code from many disciplines. The Prematurity-Related Ventilatory Control (Pre-Vent) observational multicenter prospective study collected bedside monitor data from > 700 extremely preterm infants to identify physiologic features that predict respiratory outcomes. We calculated a subset of 33 HCTSA features on > 7M 10-minute windows of oxygen saturation (SPO2) and heart rate (HR) from the Pre-Vent cohort to quantify predictive performance. This subset included representatives previously identified using unsupervised clustering on > 3500 HCTSA algorithms. Performance of each feature was measured by individual area under the receiver operating curve (AUC) at various days of life and binary respiratory outcomes. These were compared to optimal PreVent physiologic predictor IH90 DPE, the duration per event of intermittent hypoxemia events with threshold of 90%.

Main Results

The top HCTSA features were from a cluster of algorithms associated with the autocorrelation of SPO2 time series and identified low frequency patterns of desaturation as high risk. These features had comparable performance to and were highly correlated with IH90_DPE but perhaps measure the physiologic status of an infant in a more robust way that warrants further investigation. The top HR HCTSA features were symbolic transformation measures that had previously been identified as strong predictors of neonatal mortality. HR metrics were only important predictors at early days of life which was likely due to the larger proportion of infants whose outcome was death by any cause. A simple HCTSA model using 3 top features outperformed IH90_DPE at day of life 7 (.778 versus .729) but was essentially equivalent at day of life 28 (.849 versus .850). These results validated the utility of a representative HCTSA approach but also provides additional evidence supporting IH90_DPE as an optimal predictor of respiratory outcomes.

Heart rate patterns predicting cerebral palsy in preterm infants

BACKGROUND

Heart rate (HR) patterns can inform on central nervous system dysfunction. We previously used highly comparative time series analysis (HCTSA) to identify HR patterns predicting mortality among patients in the neonatal intensive care unit (NICU) and now use this methodology to discover patterns predicting cerebral palsy (CP) in preterm infants.

METHOD

We studied NICU patients <37 weeks' gestation with archived every-2-s HR data throughout the NICU stay and with or without later diagnosis of CP (n = 57 CP and 1119 no CP). We performed HCTSA of >2000 HR metrics and identified 24 metrics analyzed on HR data from two 7-day periods: week 1 and 37 weeks' postmenstrual age (week 1, week 37). Multivariate modeling was used to optimize a parsimonious prediction model.

RESULTS

Week 1 HR metrics with maximum AUC for CP prediction reflected low variability, including "RobustSD" (AUC 0.826; 0.772-0.870). At week 37, high values of a novel HR metric, "LongSD3," the cubed value of the difference in HR values 100 s apart, were added to week 1 HR metrics for CP prediction. A combined birthweight + early and late HR model had AUC 0.853 (0.805-0.892).

CONCLUSIONS

Using HCTSA, we discovered novel HR metrics and created a parsimonious model for CP prediction in preterm NICU patients.

IMPACT

We discovered new heart rate characteristics predicting CP in preterm infants. Using every-2-s HR from two 7-day periods and highly comparative time series analysis, we found a measure of low variability HR week 1 after birth and a pattern of recurrent acceleration in HR at term corrected age that predicted CP. Combined clinical and early and late HR features had AUC 0.853 for CP prediction.

Continuous ECG monitoring should be the heart of bedside AI-based predictive analytics monitoring for early detection of clinical deterioration

The idea that we can detect subacute potentially catastrophic illness earlier by using statistical models trained on clinical data is now well-established. We review evidence that supports the role of continuous cardiorespiratory monitoring in these predictive analytics monitoring tools. In particular, we review how continuous ECG monitoring reflects the patient and not the clinician, is less likely to be biased, is unaffected by changes in practice patterns, captures signatures of illnesses that are interpretable by clinicians, and is an underappreciated and underutilized source of detailed information for new mathematical methods to reveal.

The principles of whole-hospital predictive analytics monitoring for clinical medicine originated in the neonatal ICU

In 2011, a multicenter group spearheaded at the University of Virginia demonstrated reduced mortality from real-time continuous cardiorespiratory monitoring in the neonatal ICU using what we now call Artificial Intelligence, Big Data, and Machine Learning. The large, randomized heart rate characteristics trial made real, for the first time that we know of, the promise that early detection of illness would allow earlier and more effective intervention and improved patient outcomes. Currently, though, we hear as much of failures as we do of successes in the rapidly growing field of predictive analytics monitoring that has followed. This Perspective aims to describe the principles of how we developed heart rate characteristics monitoring for neonatal sepsis and then applied them throughout adult ICU and hospital medicine. It primarily reflects the work since the 1990s of the University of Virginia group: the theme is that sudden and catastrophic deteriorations can be preceded by subclinical but measurable physiological changes apparent in the continuous cardiorespiratory monitoring and electronic health record.