iCVS-Inferring Cardio-Vascular hidden States from physiological signals available at the bedside.

Intensive care medicine is complex and resource-demanding. A critical and common challenge lies in inferring the underlying physiological state of a patient from partially observed data. Specifically for the cardiovascular system, clinicians use observables such as heart rate, arterial and venous blood pressures, as well as findings from the physical examination and ancillary tests to formulate a mental model and estimate hidden variables such as cardiac output, vascular resistance, filling pressures and volumes, and autonomic tone. Then, they use this mental model to derive the causes for instability and choose appropriate interventions. Not only this is a very hard problem due to the nature of the signals, but it also requires expertise and a clinician's ongoing presence at the bedside. Clinical decision support tools based on mechanistic dynamical models offer an appealing solution due to their inherent explainability, corollaries to the clinical mental process, and predictive power. With a translational motivation in mind, we developed iCVS: a simple, with high explanatory power, dynamical mechanistic model to infer hidden cardiovascular states. Full model estimation requires no prior assumptions on physiological parameters except age and weight, and the only inputs are arterial and venous pressure waveforms. iCVS also considers autonomic and non-autonomic modulations. To gain more information without increasing model complexity, both slow and fast timescales of the blood pressure traces are exploited, while the main inference and dynamic evolution are at the longer, clinically relevant, timescale of minutes. iCVS is designed to allow bedside deployment at pediatric and adult intensive care units and for retrospective investigation of cardiovascular mechanisms underlying instability. In this paper, we describe iCVS and inference system in detail, and using a dataset of critically-ill children, we provide initial indications to its ability to identify bleeding, distributive states, and cardiac dysfunction, in isolation and in combination.

Development and validation of a multivariable prediction model in pediatric liver transplant patients for predicting intensive care unit length of stay.

BACKGROUND: Liver transplantation is the life-saving treatment for many end-stage pediatric liver diseases. The perioperative course, including surgical and anesthetic factors, have an important influence on the trajectory of this high-risk population. Given the complexity and variability of the immediate postoperative course, there would be utility in identifying risk factors that allow prediction of adverse outcomes and intensive care unit trajectories. AIMS: The aim of this study was to develop and validate a risk prediction model of prolonged intensive care unit length of stay in the pediatric liver transplant population. METHODS: This is a retrospective analysis of consecutive pediatric isolated liver transplant recipients at a single institution between April 1, 2013 and April 30, 2020. All patients under the age of 18 years receiving a liver transplant were included in the study (n = 186). The primary outcome was intensive care unit length of stay greater than 7 days. RESULTS: Recipient and donor characteristics were used to develop a multivariable logistic regression model. A total of 186 patients were included in the study. Using multivariable logistic regression, we found that age < 12 months (odds ratio 4.02, 95% confidence interval 1.20-13.51, p = .024), metabolic or cholestatic disease (odds ratio 2.66, 95% confidence interval 1.01-7.07, p = .049), 30-day pretransplant hospital admission (odds ratio 8.59, 95% confidence interval 2.27-32.54, p = .002), intraoperative red blood cells transfusion >40 mL/kg (odds ratio 3.32, 95% confidence interval 1.12-9.81, p = .030), posttransplant return to the operating room (odds ratio 11.45, 95% confidence interval 3.04-43.16, p = .004), and major postoperative respiratory event (odds ratio 32.14, 95% confidence interval 3.00-343.90, p < .001) were associated with prolonged intensive care unit length of stay. The model demonstrates a good discriminative ability with an area under the receiver operative curve of 0.888 (95% confidence interval, 0.824-0.951). CONCLUSIONS: We develop and validate a model to predict prolonged intensive care unit length of stay in pediatric liver transplant patients using risk factors from all phases of the perioperative period.

AnnoDash, a clinical terminology annotation dashboard.

Background: Standard ontologies are critical for interoperability and multisite analyses of health data. Nevertheless, mapping concepts to ontologies is often done with generic tools and is labor-intensive. Contextualizing candidate concepts within source data is also done in an ad hoc manner. Methods and Results: We present AnnoDash, a flexible dashboard to support annotation of concepts with terms from a given ontology. Text-based similarity is used to identify likely matches, and large language models are used to improve ontology ranking. A convenient interface is provided to visualize observations associated with a concept, supporting the disambiguation of vague concept descriptions. Time-series plots contrast the concept with known clinical measurements. We evaluated the dashboard qualitatively against several ontologies (SNOMED CT, LOINC, etc.) by using MIMIC-IV measurements. The dashboard is web-based and step-by-step instructions for deployment are provided, simplifying usage for nontechnical audiences. The modular code structure enables users to extend upon components, including improving similarity scoring, constructing new plots, or configuring new ontologies. Conclusion: AnnoDash, an improved clinical terminology annotation tool, can facilitate data harmonizing by promoting mapping of clinical data. AnnoDash is freely available at https://github.com/justin13601/AnnoDash (https://doi.org/10.5281/zenodo.8043943).

The truth Hertz-synchronization of electroencephalogram signals with physiological waveforms recorded in an intensive care unit.

Objective.The ability to synchronize continuous electroencephalogram (cEEG) signals with physiological waveforms such as electrocardiogram (ECG), invasive pressures, photoplethysmography and other signals can provide meaningful insights regarding coupling between brain activity and other physiological subsystems. Aligning these datasets is a particularly challenging problem because device clocks handle time differently and synchronization protocols may be undocumented or proprietary.Approach.We used an ensemble-based model to detect the timestamps of heartbeat artefacts from ECG waveforms recorded from inpatient bedside monitors and from cEEG signals acquired using a different device. Vectors of inter-beat intervals were matched between both datasets and robust linear regression was applied to measure the relative time offset between the two datasets as a function of time.Main Results.The timing error between the two unsynchronized datasets ranged between -84 s and +33 s (mean 0.77 s, median 4.31 s, IQR25-4.79 s, IQR75 11.38s). Application of our method improved the relative alignment to within ± 5ms for more than 61% of the dataset. The mean clock drift between the two datasets was 418.3 parts per million (ppm) (median 414.6 ppm, IQR25 411.0 ppm, IQR75 425.6 ppm). A signal quality index was generated that described the quality of alignment for each cEEG study as a function of time.Significance.We developed and tested a method to retrospectively time-align two clinical waveform datasets acquired from different devices using a common signal. The method was applied to 33,911h of signals collected in a paediatric critical care unit over six years, demonstrating that the method can be applied to long-term recordings collected under clinical conditions. The method can account for unknown clock drift rates and the presence of discontinuities caused by clock resynchronization events.

Pitfalls and Possibilities of Ventricular Assist Device Support in Congenitally Corrected Transposition of the Great Arteries in Children.

Congenitally corrected transposition of the great arteries (ccTGAs) represents a complex form of congenital heart disease that is associated with several cardiac complications. Herein is a case series of three children with ccTGA and ventricular assist device (VAD) inserted for systemic right ventricle failure at a single institution. All patients remained hemodynamically stable postimplant and were successfully discharged from the intensive care unit to undergo postoperative rehabilitation. All three patients received an orthotopic heart transplant with uneventful posttransplant courses. This case series provides insight into the medical management and technical feasibility of VAD support in children with ccTGA with end-stage heart failure.

Beyond Conventional Hemodynamic Monitoring-Monitoring to Improve Our Understanding of Disease Process and Interventions.

Monitoring the hemodynamic state of patients is a hallmark of any intensive care environment. However, no single monitoring strategy can provide all the necessary data to paint the entire picture of the state of a patient; each monitor has strengths and weaknesses, advantages, and limitations. We review the currently available hemodynamic monitors used in pediatric critical care units using a clinical scenario. This provides the reader with a construct to understand the progression from basic to more advanced monitoring modalities and how they serve to inform the practitioner at the bedside.

Making machine learning matter to clinicians: model actionability in medical decision-making.

Machine learning (ML) has the potential to transform patient care and outcomes. However, there are important differences between measuring the performance of ML models in silico and usefulness at the point of care. One lens to use to evaluate models during early development is actionability, which is currently undervalued. We propose a metric for actionability intended to be used before the evaluation of calibration and ultimately decision curve analysis and calculation of net benefit. Our metric should be viewed as part of an overarching effort to increase the number of pragmatic tools that identify a model's possible clinical impacts.

Preoperative Management of Neonates With Congenital Heart Disease.

Clinicians caring for neonates with congenital heart disease encounter challenges in clinical care as these infants await surgery or are evaluated for further potential interventions. The newborn with heart disease can present with significant pathophysiologic heterogeneity and therefore requires a personalized therapeutic management plan. However, this complex field of neonatal-cardiac hemodynamics can be simplified. We explore some of these clinical quandaries and include specific sections reviewing the anatomic challenges in these patients. We propose this to serve as a primer focusing on the hemodynamics and therapeutic strategies for the preoperative neonate with systolic dysfunction, diastolic dysfunction, excessive pulmonary blood flow, obstructed pulmonary blood flow, obstructed systemic blood flow, transposition physiology, and single ventricle physiology.

Timing errors and temporal uncertainty in clinical databases-A narrative review.

A firm concept of time is essential for establishing causality in a clinical setting. Review of critical incidents and generation of study hypotheses require a robust understanding of the sequence of events but conducting such work can be problematic when timestamps are recorded by independent and unsynchronized clocks. Most clinical models implicitly assume that timestamps have been measured accurately and precisely, but this custom will need to be re-evaluated if our algorithms and models are to make meaningful use of higher frequency physiological data sources. In this narrative review we explore factors that can result in timestamps being erroneously recorded in a clinical setting, with particular focus on systems that may be present in a critical care unit. We discuss how clocks, medical devices, data storage systems, algorithmic effects, human factors, and other external systems may affect the accuracy and precision of recorded timestamps. The concept of temporal uncertainty is introduced, and a holistic approach to timing accuracy, precision, and uncertainty is proposed. This quantitative approach to modeling temporal uncertainty provides a basis to achieve enhanced model generalizability and improved analytical outcomes.

Analyzing Continuous Physiologic Data to Find Hemodynamic Signatures Associated With New Brain Injury After Congenital Heart Surgery.

Continuous data capture technology is becoming more common. Establishing analytic approaches for continuous data could aid in understanding the relationship between physiology and clinical outcomes. OBJECTIVES: Our objective was to design a retrospective analysis for continuous physiologic measurements and their relationship with new brain injury over time after cardiac surgery. DESIGN SETTING AND PARTICIPANTS: Retrospective cohort study in the Cardiac Critical Care Unit at the Hospital for Sick Children in patients after repair of transposition of the great arteries (TGA) or single ventricle (SV) lesions. MAIN OUTCOMES AND MEASURES: Continuously acquired physiologic measurements for up to 72 hours after cardiac surgery were analyzed for association with new brain injury by MRI. Distributions of heart rate (HR), systolic blood pressure (BP), and oxygen saturation (Spo2) for SV and TGA were analyzed graphically and with descriptive statistics over postoperative time for data-driven variable selection. Mixed-effects regression analyses characterized relationships between HR, BP, and Spo2 and new brain injury over time while accounting for variation between patients, measurement heterogeneity, and missingness. RESULTS: Seventy-seven patients (60 TGA; 17 SV) were included. New brain injury was seen in 26 (34%). In SV patients, with and without new brain injury, respectively, in the first 24 hours after cardiac surgery, the median (interquartile range) HR was 172.0 beats/min (bpm) (169.7-176.0 bpm) versus 159.6 bpm (145.0-167.0 bpm); systolic BP 74.8 (67.9-78.5 mm Hg) versus 68.9 mm Hg (61.6-70.9 mm Hg). Higher postoperative HR (parameter estimate, 19.4; 95% CI, 7.8-31; p = 0.003 and BP, 8.6; 1.3-15.8; p = 0.024) were associated with new brain injury in SV patients. The strength of this relationship decreased with time. CONCLUSIONS AND RELEVANCE: Retrospective analysis of continuous physiologic measurements can provide insight into changes in postoperative physiology over time and their relationship with new brain injury. This technique could be applied to assess relationships between physiologic data and many patient interventions or outcomes.

An integration engineering framework for machine learning in healthcare.

Background and Objectives: Machine Learning offers opportunities to improve patient outcomes, team performance, and reduce healthcare costs. Yet only a small fraction of all Machine Learning models for health care have been successfully integrated into the clinical space. There are no current guidelines for clinical model integration, leading to waste, unnecessary costs, patient harm, and decreases in efficiency when improperly implemented. Systems engineering is widely used in industry to achieve an integrated system of systems through an interprofessional collaborative approach to system design, development, and integration. We propose a framework based on systems engineering to guide the development and integration of Machine Learning models in healthcare. Methods: Applied systems engineering, software engineering and health care Machine Learning software development practices were reviewed and critically appraised to establish an understanding of limitations and challenges within these domains. Principles of systems engineering were used to develop solutions to address the identified problems. The framework was then harmonized with the Machine Learning software development process to create a systems engineering-based Machine Learning software development approach in the healthcare domain. Results: We present an integration framework for healthcare Artificial Intelligence that considers the entirety of this system of systems. Our proposed framework utilizes a combined software and integration engineering approach and consists of four phases: (1) Inception, (2) Preparation, (3) Development, and (4) Integration. During each phase, we present specific elements for consideration in each of the three domains of integration: The Human, The Technical System, and The Environment. There are also elements that are considered in the interactions between these domains. Conclusion: Clinical models are technical systems that need to be integrated into the existing system of systems in health care. A systems engineering approach to integration ensures appropriate elements are considered at each stage of model design to facilitate model integration. Our proposed framework is based on principles of systems engineering and can serve as a guide for model development, increasing the likelihood of successful Machine Learning translation and integration.

Ignorance Isn't Bliss: We Must Close the Machine Learning Knowledge Gap in Pediatric Critical Care.

Pediatric intensivists are bombarded with more patient data than ever before. Integration and interpretation of data from patient monitors and the electronic health record (EHR) can be cognitively expensive in a manner that results in delayed or suboptimal medical decision making and patient harm. Machine learning (ML) can be used to facilitate insights from healthcare data and has been successfully applied to pediatric critical care data with that intent. However, many pediatric critical care medicine (PCCM) trainees and clinicians lack an understanding of foundational ML principles. This presents a major problem for the field. We outline the reasons why in this perspective and provide a roadmap for competency-based ML education for PCCM trainees and other stakeholders.

Successful Pediatric Mechanical Thrombectomy for HeartMate 3-Related Intracranial Thromboembolism.

Thromboembolic events post left ventricular assist devices (LVAD) implantation remain a major cause of morbidity and mortality. Mechanical thrombectomy for the treatment of pediatric intracranial thromboembolic events have been reported in LVADs, but never following HeartMate 3 (HM3) implantation. We present the case of an 8-year-old, 26.5 kg male with dilated cardiomyopathy and decompensated heart failure who presented with extensive intracranial thromboembolism in the early postoperative period following HM3 implantation and underwent successful mechanical thrombectomy with a favorable neurological outcome.

Care of the critically ill neonate with hypoxemic respiratory failure and acute pulmonary hypertension: framework for practice based on consensus opinion of neonatal hemodynamics working group.

Circulatory transition after birth presents a critical period whereby the pulmonary vascular bed and right ventricle must adapt to rapidly changing loading conditions. Failure of postnatal transition may present as hypoxemic respiratory failure, with disordered pulmonary and systemic blood flow. In this review, we present the biological and clinical contributors to pathophysiology and present a management framework.

Relieving bronchial compression due to cardiomegaly: The role of aortopexy when left ventricular assist device support just is not enough.

Enlarged cardiac structures, especially those on left side have the potential to cause airway compression in pediatric patients with chronic heart failure, owing to their proximity to and impact on the trachea-bronchial tree. Ventricular assist devices are effective in decompressing such hearts thereby alleviating airway problems. Aortopexy serves as an effective airway decompressive measure in cases with persistent airway compression despite effective cardiac decompression by ventricular assist devices. We report a case of 1-year-old male patient with dilated cardiomyopathy in whom airway compression persisted despite ventricular assist device implantation. Aortopexy was effective in relieving airway compression allowing for subsequent extubation and successful heart transplantation.

Hemodynamic Patterns Before Inhospital Cardiac Arrest in Critically Ill Children: An Exploratory Study.

To characterize prearrest hemodynamic trajectories of children suffering inhospital cardiac arrest. DESIGN: Exploratory retrospective analysis of arterial blood pressure and electrocardiogram waveforms. SETTING: PICU and cardiac critical care unit in a tertiary-care children's hospital. PATIENTS: Twenty-seven children with invasive blood pressure monitoring who suffered a total of 31 inhospital cardiac arrest events between June 2017 and June 2019. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We assessed changes in cardiac output, systemic vascular resistance, stroke volume, and heart rate derived from arterial blood pressure waveforms using three previously described estimation methods. We observed substantial prearrest drops in cardiac output (population median declines of 65-84% depending on estimation method) in all patients in the 10 minutes preceding inhospital cardiac arrest. Most patients' mean arterial blood pressure also decreased, but this was not universal. We identified three hemodynamic patterns preceding inhospital cardiac arrest: subacute pulseless arrest (n = 18), acute pulseless arrest (n = 7), and bradycardic arrest (n = 6). Acute pulseless arrest events decompensated within seconds, whereas bradycardic and subacute pulseless arrest events deteriorated over several minutes. In the subacute and acute pulseless arrest groups, decreases in cardiac output were primarily due to declines in stroke volume, whereas in the bradycardic group, the decreases were primarily due to declines in heart rate. CONCLUSIONS: Critically ill children exhibit distinct physiologic behaviors prior to inhospital cardiac arrest. All events showed substantial declines in cardiac output shortly before inhospital cardiac arrest. We describe three distinct prearrest patterns with varying rates of decline and varying contributions of heart rate and stroke volume changes to the fall in cardiac output. Our findings suggest that monitoring changes in arterial blood pressure waveform-derived heart rate, pulse pressure, cardiac output, and systemic vascular resistance estimates could improve early detection of inhospital cardiac arrest by up to several minutes. Further study is necessary to verify the patterns witnessed in our cohort as a step toward patient rather than provider-centered definitions of inhospital cardiac arrest.