Vital sign metrics of VLBW infants in three NICUs: implications for predictive algorithms

Abstract

Background: Continuous heart rate (HR) and oxygenation (SpO 2 ) metrics can be useful for predicting adverse events in very low birth weight (VLBW) infants. To optimize the utility of these tools, inter-site variability must be taken into account. Methods:For VLBW infants at three NICUs, we analyzed the mean, standard deviation, skewness, kurtosis, and cross-correlation of electrocardiogram HR, pulse oximeter pulse rate, and SpO 2 . The number and durations of bradycardia and desaturation events were also measured. Twenty-two metrics were calculated hourly, and mean daily values were compared between sites. Results:We analyzed data from 1,168 VLBW infants from birth through day 42 (35,238 infant days). HR and SpO 2 metrics were similar at the three NICUs, with mean HR rising by ~10 beats/minute over the first two weeks and mean SpO 2 remaining stable around 94% over time. The number of bradycardia events was higher at one site, and the duration of desaturations was longer at another site.Users may view, print, copy, and download text and data-mine the content in such documents, for the purposes of academic research, subject always to the full Conditions of use:http://www.nature.com/authors/editorial_policies/license.html#terms

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

“…External validation will be important because our findings of patterns in vital signs measurements prior to neonatal death might reflect care practices at our hospital. We note, though, the similarity of vital signs measurements at our hospital to those at two others, 45 a finding that is reassuring with regard to the general nature of these results. We found that only 3555 of the 4998 algorithms consistently returned non-null results, and we note that other data sets from other sources might fare differently.…”

“…We did not analyze pre-term infants separately from term infants in this work, though we know that heart rate and SpO2 time series characteristics depend on both gestational age and postconceptual age. For example, the variabilities of heart rate and SpO2 rise with day of age, 45,46 and it is possible that highly-comparative time series analysis of pre-term infants might return different results from term infants. As it stands, there were many more term infants than preterm, but the latter represented more of the time series data.…”

Cardiorespiratory signature of neonatal sepsis: Development and validation of prediction models in 3 NICUs

“…We found HR-SpO 2 cross-correlation to be the best individual feature to discriminate LOS vs. sepsis-ruled out events. 5 We also compared vital signs across our three collaborating sites and found clinically trivial though statistically significant differences in HR and SpO2 35 .…”

Application and potential of artificial intelligence in neonatal medicine

“…This can lead to early recognition of disease states, sometimes hours before they would become clinically apparent [10]. ML algorithms can outperform HCPs in fulfilling this task, recognising the disease and supporting clinical decision making [7,[10][11][12].…”

“…To evaluate how the model interacts with data throughout the infant's entire NICU stay, it should be subjected to continuous performance analysis, evaluating its output at every single timepoint [7]. Moreover, ML algorithms should be compared with current clinical tools in diagnostic accuracy, be subjected to internal validation and undergo generalisability testing with external validation [12].…”