Clinical correlates of a high cardiorespiratory risk score for very low birth weight infants.

BACKGROUND: A pulse oximetry warning system (POWS) to analyze heart rate and oxygen saturation data and predict risk of sepsis was developed for very low birth weight (VLBW) infants. METHODS: We determined the clinical correlates and positive predictive value (PPV) of a high POWS score in VLBW infants. In a two-NICU retrospective study, we identified times when POWS increased above 6 (POWS spike). We selected an equal number of control times, matched for gestational and chronologic age. We reviewed records for infection and non-infection events around POWS spikes and control times. We calculated the frequencies and PPV of a POWS spike for infection or another significant event. RESULTS: We reviewed 111 POWS spike times and 111 control times. Days near POWS spikes were more likely to have clinical events than control days (77% vs 50%). A POWS spike had 52% PPV for suspected or confirmed infection and 77% for any clinically significant event. Respiratory deterioration occurred near more POWS spike times than control times (34% vs 18%). CONCLUSIONS: In a retrospective cohort, infection and respiratory deterioration were common clinical correlations of a POWS spike. POWS had a high PPV for significant clinical events with or without infection. IMPACT: There are significant gaps in understanding the best approach to implementing continuous sepsis prediction models so that clinicians can best respond to early signals of deterioration. Infection and respiratory deterioration were common clinical events identified at the time of a high predictive model score. Understanding the clinical correlates of a high-risk early warning score will inform future implementation efforts.

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.

Apnea, Intermittent Hypoxemia, and Bradycardia Events Predict Late-Onset Sepsis in Infants Born Extremely Preterm.

OBJECTIVE: The objective of this study was to examine the association of cardiorespiratory events, including apnea, periodic breathing, intermittent hypoxemia (IH), and bradycardia, with late-onset sepsis for extremely preterm infants (<29 weeks of gestational age) on vs off invasive mechanical ventilation. STUDY DESIGN: This is a retrospective analysis of data from infants enrolled in Pre-Vent (ClinicalTrials.gov identifier NCT03174301), an observational study in 5 level IV neonatal intensive care units. Clinical data were analyzed for 737 infants (mean gestational age: 26.4 weeks, SD 1.71). Monitoring data were available and analyzed for 719 infants (47 512 patient-days); of whom, 109 had 123 sepsis events. Using continuous monitoring data, we quantified apnea, periodic breathing, bradycardia, and IH. We analyzed the relationships between these daily measures and late-onset sepsis (positive blood culture >72 hours after birth and ≥5-day antibiotics). RESULTS: For infants not on a ventilator, apnea, periodic breathing, and bradycardia increased before sepsis diagnosis. During times on a ventilator, increased sepsis risk was associated with longer events with oxygen saturation <80% (IH80) and more bradycardia events before sepsis. IH events were associated with higher sepsis risk but did not dynamically increase before sepsis, regardless of ventilator status. A multivariable model including postmenstrual age, cardiorespiratory variables (apnea, periodic breathing, IH80, and bradycardia), and ventilator status predicted sepsis with an area under the receiver operator characteristic curve of 0.783. CONCLUSION: We identified cardiorespiratory signatures of late-onset sepsis. Longer IH events were associated with increased sepsis risk but did not change temporally near diagnosis. Increases in bradycardia, apnea, and periodic breathing preceded the clinical diagnosis of sepsis.

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.

Apnea, Intermittent Hypoxemia, and Bradycardia Events Predict Late-Onset Sepsis in Extremely Preterm Infants.

Objectives: Detection of changes in cardiorespiratory events, including apnea, periodic breathing, intermittent hypoxemia (IH), and bradycardia, may facilitate earlier detection of sepsis. Our objective was to examine the association of cardiorespiratory events with late-onset sepsis for extremely preterm infants (<29 weeks' gestational age (GA)) on versus off invasive mechanical ventilation. Study Design: Retrospective analysis of data from infants enrolled in Pre-Vent (ClinicalTrials.gov identifier NCT03174301), an observational study in five level IV neonatal intensive care units. Clinical data were analyzed for 737 infants (mean GA 26.4w, SD 1.71). Monitoring data were available and analyzed for 719 infants (47,512 patient-days), of whom 109 had 123 sepsis events. Using continuous monitoring data, we quantified apnea, periodic breathing, bradycardia, and IH. We analyzed the relationships between these daily measures and late-onset sepsis (positive blood culture >72h after birth and ≥5d antibiotics). Results: For infants not on a ventilator, apnea, periodic breathing, and bradycardia increased before sepsis diagnosis. During times on a ventilator, increased sepsis risk was associated with longer IH80 events and more bradycardia events before sepsis. IH events were associated with higher sepsis risk, but did not dynamically increase before sepsis, regardless of ventilator status. A multivariable model predicted sepsis with an AUC of 0.783. Conclusion: We identified cardiorespiratory signatures of late-onset sepsis. Longer IH events were associated with increased sepsis risk but did not change temporally near diagnosis. Increases in bradycardia, apnea, and periodic breathing preceded the clinical diagnosis of sepsis.

Maturation of cardioventilatory physiological trajectories in extremely preterm infants.

BACKGROUND: In extremely preterm infants, persistence of cardioventilatory events is associated with long-term morbidity. Therefore, the objective was to characterize physiologic growth curves of apnea, periodic breathing, intermittent hypoxemia, and bradycardia in extremely preterm infants during the first few months of life. METHODS: The Prematurity-Related Ventilatory Control study included 717 preterm infants <29 weeks gestation. Waveforms were downloaded from bedside monitors with a novel sharing analytics strategy utilized to run software locally, with summary data sent to the Data Coordinating Center for compilation. RESULTS: Apnea, periodic breathing, and intermittent hypoxemia events rose from day 3 of life then fell to near-resolution by 8-12 weeks of age. Apnea/intermittent hypoxemia were inversely correlated with gestational age, peaking at 3-4 weeks of age. Periodic breathing was positively correlated with gestational age peaking at 31-33 weeks postmenstrual age. Females had more periodic breathing but less intermittent hypoxemia/bradycardia. White infants had more apnea/periodic breathing/intermittent hypoxemia. Infants never receiving mechanical ventilation followed similar postnatal trajectories but with less apnea and intermittent hypoxemia, and more periodic breathing. CONCLUSIONS: Cardioventilatory events peak during the first month of life but the actual postnatal trajectory is dependent on the type of event, race, sex and use of mechanical ventilation. IMPACT: Physiologic curves of cardiorespiratory events in extremely preterm-born infants offer (1) objective measures to assess individual patient courses and (2) guides for research into control of ventilation, biomarkers and outcomes. Presented are updated maturational trajectories of apnea, periodic breathing, intermittent hypoxemia, and bradycardia in 717 infants born <29 weeks gestation from the multi-site NHLBI-funded Pre-Vent study. Cardioventilatory events peak during the first month of life but the actual postnatal trajectory is dependent on the type of event, race, sex and use of mechanical ventilation. Different time courses for apnea and periodic breathing suggest different maturational mechanisms.

Cardiorespiratory Monitoring Data to Predict Respiratory Outcomes in Extremely Preterm Infants.

Rationale: Immature control of breathing is associated with apnea, periodic breathing, intermittent hypoxemia, and bradycardia in extremely preterm infants. However, it is not clear if such events independently predict worse respiratory outcome. Objectives: To determine if analysis of cardiorespiratory monitoring data can predict unfavorable respiratory outcomes at 40 weeks postmenstrual age (PMA) and other outcomes, such as bronchopulmonary dysplasia at 36 weeks PMA. Methods: The Prematurity-related Ventilatory Control (Pre-Vent) study was an observational multicenter prospective cohort study including infants born at <29 weeks of gestation with continuous cardiorespiratory monitoring. The primary outcome was either "favorable" (alive and previously discharged or inpatient and off respiratory medications/O2/support at 40 wk PMA) or "unfavorable" (either deceased or inpatient/previously discharged on respiratory medications/O2/support at 40 wk PMA). Measurements and Main Results: A total of 717 infants were evaluated (median birth weight, 850 g; gestation, 26.4 wk), 53.7% of whom had a favorable outcome and 46.3% of whom had an unfavorable outcome. Physiologic data predicted unfavorable outcome, with accuracy improving with advancing age (area under the curve, 0.79 at Day 7, 0.85 at Day 28 and 32 wk PMA). The physiologic variable that contributed most to prediction was intermittent hypoxemia with oxygen saturation as measured by pulse oximetry <90%. Models with clinical data alone or combining physiologic and clinical data also had good accuracy, with areas under the curve of 0.84-0.85 at Days 7 and 14 and 0.86-0.88 at Day 28 and 32 weeks PMA. Intermittent hypoxemia with oxygen saturation as measured by pulse oximetry <80% was the major physiologic predictor of severe bronchopulmonary dysplasia and death or mechanical ventilation at 40 weeks PMA. Conclusions: Physiologic data are independently associated with unfavorable respiratory outcome in extremely preterm infants.

A novel predictive analytics score reflecting accumulating disease burden-an investigation of the cumulative CoMET score.

OBJECTIVE: Predictive analytics tools variably take into account data from the electronic medical record, lab tests, nursing charted vital signs and continuous cardiorespiratory monitoring to deliver an instantaneous prediction of patient risk or instability. Few, if any, of these tools reflect the risk to a patient accumulated over the course of an entire hospital stay. APPROACH: We have expanded on our instantaneous CoMET predictive analytics score to generate the cumulative CoMET score (cCoMET), which sums all of the instantaneous CoMET scores throughout a hospital admission relative to a baseline expected risk unique to that patient. MAIN RESULTS: We have shown that higher cCoMET scores predict mortality, but not length of stay, and that higher baseline CoMET scores predict higher cCoMET scores at discharge/death. cCoMET scores were higher in males in our cohort, and added information to the final CoMET when it came to the prediction of death. SIGNIFICANCE: We have shown that the inclusion of all repeated measures of risk estimation performed throughout a patients hospital stay adds information to instantaneous predictive analytics, and could improve the ability of clinicians to predict deterioration, and improve patient outcomes in so doing.

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

BACKGROUND: Heart rate characteristics aid early detection of late-onset sepsis (LOS), but respiratory data contain additional signatures of illness due to infection. Predictive models using cardiorespiratory data may improve early sepsis detection. We hypothesized that heart rate (HR) and oxygenation (SpO2) data contain signatures that improve sepsis risk prediction over HR or demographics alone. METHODS: We analyzed cardiorespiratory data from very low birth weight (VLBW, <1500 g) infants admitted to three NICUs. We developed and externally validated four machine learning models to predict LOS using features calculated every 10 m: mean, standard deviation, skewness, kurtosis of HR and SpO2, and cross-correlation. We compared feature importance, discrimination, calibration, and dynamic prediction across models and cohorts. We built models of demographics and HR or SpO2 features alone for comparison with HR-SpO2 models. RESULTS: Performance, feature importance, and calibration were similar among modeling methods. All models had favorable external validation performance. The HR-SpO2 model performed better than models using either HR or SpO2 alone. Demographics improved the discrimination of all physiologic data models but dampened dynamic performance. CONCLUSIONS: Cardiorespiratory signatures detect LOS in VLBW infants at 3 NICUs. Demographics risk-stratify, but predictive modeling with both HR and SpO2 features provides the best dynamic risk prediction. IMPACT: Heart rate characteristics aid early detection of late-onset sepsis, but respiratory data contain signatures of illness due to infection. Predictive models using both heart rate and respiratory data may improve early sepsis detection. A cardiorespiratory early warning score, analyzing heart rate from electrocardiogram or pulse oximetry with SpO2, predicts late-onset sepsis within 24 h across multiple NICUs and detects sepsis better than heart rate characteristics or demographics alone. Demographics risk-stratify, but predictive modeling with both HR and SpO2 features provides the best dynamic risk prediction. The results increase understanding of physiologic signatures of neonatal sepsis.

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.

Signatures of illness in children requiring unplanned intubation in the pediatric intensive care unit: A retrospective cohort machine-learning study.

Acute respiratory failure requiring the initiation of invasive mechanical ventilation remains commonplace in the pediatric intensive care unit (PICU). Early recognition of patients at risk for respiratory failure may provide clinicians with the opportunity to intervene and potentially improve outcomes. Through the development of a random forest model to identify patients at risk for requiring unplanned intubation, we tested the hypothesis that subtle signatures of illness are present in physiological and biochemical time series of PICU patients in the early stages of respiratory decompensation. We included 116 unplanned intubation events as recorded in the National Emergency Airway Registry for Children in 92 PICU admissions over a 29-month period at our institution. We observed that children have a physiologic signature of illness preceding unplanned intubation in the PICU. Generally, it comprises younger age, and abnormalities in electrolyte, hematologic and vital sign parameters. Additionally, given the heterogeneity of the PICU patient population, we found differences in the presentation among the major patient groups - medical, cardiac surgical, and non-cardiac surgical. At four hours prior to the event, our random forest model demonstrated an area under the receiver operating characteristic curve of 0.766 (0.738 for medical, 0.755 for cardiac surgical, and 0.797 for non-cardiac surgical patients). The multivariable statistical models that captured the physiological and biochemical dynamics leading up to the event of urgent unplanned intubation in a PICU can be repurposed for bedside risk prediction.

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

To seek new signatures of illness in heart rate and oxygen saturation vital signs from Neonatal Intensive Care Unit (NICU) patients, we implemented highly comparative time-series analysis to discover features of all-cause mortality in the next 7 days. We collected 0.5 Hz heart rate and oxygen saturation vital signs of infants in the University of Virginia NICU from 2009 to 2019. We applied 4998 algorithmic operations from 11 mathematical families to random daily 10 min segments from 5957 NICU infants, 205 of whom died. We clustered the results and selected a representative from each, and examined multivariable logistic regression models. 3555 operations were usable; 20 cluster medoids held more than 81% of the information, and a multivariable model had AUC 0.83. New algorithms outperformed others: moving threshold, successive increases, surprise, and random walk. We computed provenance of the computations and constructed a software library with links to the data. We conclude that highly comparative time-series analysis revealed new vital sign measures to identify NICU patients at the highest risk of death in the next week.

FAIRSCAPE: a Framework for FAIR and Reproducible Biomedical Analytics.

Results of computational analyses require transparent disclosure of their supporting resources, while the analyses themselves often can be very large scale and involve multiple processing steps separated in time. Evidence for the correctness of any analysis should include not only a textual description, but also a formal record of the computations which produced the result, including accessible data and software with runtime parameters, environment, and personnel involved. This article describes FAIRSCAPE, a reusable computational framework, enabling simplified access to modern scalable cloud-based components. FAIRSCAPE fully implements the FAIR data principles and extends them to provide fully FAIR Evidence, including machine-interpretable provenance of datasets, software and computations, as metadata for all computed results. The FAIRSCAPE microservices framework creates a complete Evidence Graph for every computational result, including persistent identifiers with metadata, resolvable to the software, computations, and datasets used in the computation; and stores a URI to the root of the graph in the result's metadata. An ontology for Evidence Graphs, EVI ( https://w3id.org/EVI ), supports inferential reasoning over the evidence. FAIRSCAPE can run nested or disjoint workflows and preserves provenance across them. It can run Apache Spark jobs, scripts, workflows, or user-supplied containers. All objects are assigned persistent IDs, including software. All results are annotated with FAIR metadata using the evidence graph model for access, validation, reproducibility, and re-use of archived data and software.

External validation of a novel signature of illness in continuous cardiorespiratory monitoring to detect early respiratory deterioration of ICU patients.

OBJECTIVE: The goal of predictive analytics monitoring is the early detection of patients at high risk of subacute potentially catastrophic illnesses. An excellent example of a targeted illness is respiratory failure leading to urgent unplanned intubation, where early detection might lead to interventions that improve patient outcomes. Previously, we identified signatures of this illness in the continuous cardiorespiratory monitoring data of intensive care unit (ICU) patients and devised algorithms to identify patients at rising risk. Here, we externally validated three logistic regression models to estimate the risk of emergency intubation developed in Medical and Surgical ICUs at the University of Virginia. APPROACH: We calculated the model outputs for more than 8000 patients in the University of California-San Francisco ICUs, 240 of whom underwent emergency intubation as determined by individual chart review. MAIN RESULTS: We found that the AUC of the models exceeded 0.75 in this external population, and that the risk rose appreciably over the 12 h before the event. SIGNIFICANCE: We conclude that there are generalizable physiological signatures of impending respiratory failure in the continuous cardiorespiratory monitoring data.

Physiological machine learning models for prediction of sepsis in hospitalized adults: An integrative review.

BACKGROUND: Diagnosing sepsis remains challenging. Data compiled from continuous monitoring and electronic health records allow for new opportunities to compute predictions based on machine learning techniques. There has been a lack of consensus identifying best practices for model development and validation towards early identification of sepsis. OBJECTIVE: To evaluate the modeling approach and statistical methodology of machine learning prediction models for sepsis in the adult hospital population. METHODS: PubMed, CINAHL, and Cochrane databases were searched with the Preferred Reporting Items for Systematic Reviews guided protocol development. We evaluated studies that developed or validated physiologic sepsis prediction models or implemented a model in the hospital environment. RESULTS: Fourteen studies met the inclusion criteria, and the AUROC of the prediction models ranged from 0.61 to 0.96. We found a variety of sepsis definitions, methods used for event adjudication, model parameters used, and modeling methods. Two studies tested models in clinical settings; the results suggested that patient outcomes were improved with implementation of machine learning models. CONCLUSION: Nurses have a unique perspective to offer in the development and implementation of machine learning models detecting patients at risk for sepsis. More work is needed in developing model harmonization standards and testing in clinical settings.

The critical care data exchange format: a proposed flexible data standard for combining clinical and high-frequency physiologic data in critical care.

Objective.To develop a standardized format for exchanging clinical and physiologic data generated in the intensive care unit. Our goal was to develop a format that would accommodate the data collection pipelines of various sites but would not require dataset-specific schemas or ad-hoc tools for decoding and analysis.Approach.A number of centers had independently developed solutions for storing clinical and physiologic data using Hierarchical Data Format-Version 5 (HDF5), a well-supported standard already in use in multiple other fields. These individual solutions involved design choices that made the data difficult to share despite the underlying common framework. A collaborative process was used to form the basis of a proposed standard that would allow for interoperability and data sharing with common analysis tools.Main Results.We developed the HDF5-based critical care data exchange format which stores multiparametric data in an efficient, self-describing, hierarchical structure and supports real-time streaming and compression. In addition to cardiorespiratory and laboratory data, the format can, in future, accommodate other large datasets such as imaging and genomics. We demonstated the feasibility of a standardized format by converting data from three sites as well as the MIMIC III dataset.Significance.Individual approaches to the storage of multiparametric clinical data are proliferating, representing both a duplication of effort and a missed opportunity for collaboration. Adoption of a standardized format for clinical data exchange will enable the development of a digital biobank, facilitate the external validation of machine learning models and be a powerful tool for sharing multiparametric, high frequency patient level data in multisite clinical trials. Our proposed solution focuses on supporting standardized ontologies such as LOINC allowing easy reading of data regardless of the source and in so doing provides a useful method to integrate large amounts of existing data.

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

BACKGROUND: Continuous heart rate (HR) and oxygenation (SpO2) 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 neonatal intensive care units (NICUs), we analyzed the mean, standard deviation, skewness, kurtosis, and cross-correlation of electrocardiogram HR, pulse oximeter pulse rate, and SpO2. 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 1168 VLBW infants from birth through day 42 (35,238 infant-days). HR and SpO2 metrics were similar at the three NICUs, with mean HR rising by ~10 beats/min over the first 2 weeks and mean SpO2 remaining stable ~94% over time. The number of bradycardia events was higher at one site, and the duration of desaturations was longer at another site. CONCLUSIONS: Mean HR and SpO2 were generally similar among VLBW infants at three NICUs from birth through 6 weeks of age, but bradycardia and desaturation events differed in the first 2 weeks after birth. This highlights the importance of developing predictive analytics tools at multiple sites. IMPACT: HR and SpO2 analytics can be useful for predicting adverse events in VLBW infants in the NICU, but inter-site differences must be taken into account in developing predictive algorithms. Although mean HR and SpO2 patterns were similar in VLBW infants at three NICUs, inter-site differences in the number of bradycardia events and duration of desaturation events were found. Inter-site differences in bradycardia and desaturation events among VLBW infants should be considered in the development of predictive algorithms.

Autism risk in neonatal intensive care unit patients associated with novel heart rate patterns.

BACKGROUND: Neonatal intensive care unit (NICU) patients are at increased risk for autism spectrum disorder (ASD). Autonomic nervous system aberrancy has been described in children with ASD, and we aimed to identify heart rate (HR) patterns in NICU patients associated with eventual ASD diagnosis. METHODS: This retrospective cohort study included NICU patients from 2009 to 2016 with archived HR data and follow-up beyond age 3 years. Medical records provided clinical variables and ASD diagnosis. HR data were compared in infants with and without ASD. RESULTS: Of the 2371 patients, 88 had ASD, and 689,016 h of data were analyzed. HR skewness (HRskw) was significantly different between ASD and control infants. Preterm infants at early postmenstrual ages (PMAs) had negative HRskw reflecting decelerations, which increased with maturation. From 34 to 42 weeks PMA, positive HRskw toward accelerations was higher in males with ASD. In 931 males with at least 4 days of HR data, overall ASD prevalence was 5%, whereas 11% in the top 5th HRskw percentile had ASD. CONCLUSION: High HRskw in NICU males, perhaps representing autonomic imbalance, was associated with increased ASD risk. Further study is needed to determine whether HR analysis identifies highest-risk infants who might benefit from earlier screening and therapies. IMPACT: In a large retrospective single-center cohort of NICU patients, we found that high positive skewness of heart rate toward more accelerations was significantly associated with increased risk of eventual autism spectrum disorder diagnosis in male infants but not in females. Existing literature describes differences in heart rate characteristics in children, adolescents, and adults with autism spectrum disorders, but the finding from our study in NICU infants is novel. Heart rate analysis during the NICU stay might identify, among an inherently high-risk population, those infants with especially high risk of ASD who might benefit from earlier screening and therapies.