Application of Cardiovascular Physiology to the Critically Ill Patient.

OBJECTIVES: To use the ventricular pressure-volume relationship and time-varying elastance model to provide a foundation for understanding cardiovascular physiology and pathophysiology, interpreting advanced hemodynamic monitoring, and for illustrating the physiologic basis and hemodynamic effects of therapeutic interventions. We will build on this foundation by using a cardiovascular simulator to illustrate the application of these principles in the care of patients with severe sepsis, cardiogenic shock, and acute mechanical circulatory support. DATA SOURCES: Publications relevant to the discussion of the time-varying elastance model, cardiogenic shock, and sepsis were retrieved from MEDLINE. Supporting evidence was also retrieved from MEDLINE when indicated. STUDY SELECTION, DATA EXTRACTION, AND SYNTHESIS: Data from relevant publications were reviewed and applied as indicated. CONCLUSIONS: The ventricular pressure-volume relationship and time-varying elastance model provide a foundation for understanding cardiovascular physiology and pathophysiology. We have built on this foundation by using a cardiovascular simulator to illustrate the application of these important principles and have demonstrated how complex pathophysiologic abnormalities alter clinical parameters used by the clinician at the bedside.

Discovering the Clinical Relevance of Heart-Lung Interactions.

In 1978, Dr. Pinsky's scientific career became firmly directed toward understanding the deeper meaning of heart-lung interactions. This would define his focus for the next 45 yr. At the time, he and colleagues studied the effects of changes in intrathoracic pressure on left ventricular performance in humans, documenting that the primary effect of large negative swings in intrathoracic pressure was to increase left ventricular transmural ejection pressure, and thus left ventricular afterload, selectively. They concluded that large intrathoracic pressure changes directly influence cardiac performance. This fundamental observation was followed by many additional observations in both highly invasive animal studies supported by less invasive clinical studies, which showed that intrathoracic pressure-induced changes in the gradients for venous return to the heart and left ventricular ejection from the heart disproportionately affected both right ventricular and left ventricular function. The direct clinical implications of these results form the rationale for use of continuous positive airway pressure as a primary treatment of acute cardiogenic pulmonary edema and immediate endotracheal intubation for acute upper airway obstruction. These findings subsequently led to the practical use of dynamic changes in left ventricular stroke volume and the associated arterial pulse pressure during positive-pressure ventilation to identify volume responsiveness and, thus, to personalize resuscitation efforts in the treatment of acute cardiovascular insufficiency.

Use of artificial intelligence in critical care: opportunities and obstacles.

BACKGROUND: Perhaps nowhere else in the healthcare system than in the intensive care unit environment are the challenges to create useful models with direct time-critical clinical applications more relevant and the obstacles to achieving those goals more massive. Machine learning-based artificial intelligence (AI) techniques to define states and predict future events are commonplace activities of modern life. However, their penetration into acute care medicine has been slow, stuttering and uneven. Major obstacles to widespread effective application of AI approaches to the real-time care of the critically ill patient exist and need to be addressed. MAIN BODY: Clinical decision support systems (CDSSs) in acute and critical care environments support clinicians, not replace them at the bedside. As will be discussed in this review, the reasons are many and include the immaturity of AI-based systems to have situational awareness, the fundamental bias in many large databases that do not reflect the target population of patient being treated making fairness an important issue to address and technical barriers to the timely access to valid data and its display in a fashion useful for clinical workflow. The inherent "black-box" nature of many predictive algorithms and CDSS makes trustworthiness and acceptance by the medical community difficult. Logistically, collating and curating in real-time multidimensional data streams of various sources needed to inform the algorithms and ultimately display relevant clinical decisions support format that adapt to individual patient responses and signatures represent the efferent limb of these systems and is often ignored during initial validation efforts. Similarly, legal and commercial barriers to the access to many existing clinical databases limit studies to address fairness and generalizability of predictive models and management tools. CONCLUSIONS: AI-based CDSS are evolving and are here to stay. It is our obligation to be good shepherds of their use and further development.

Severity-Driven Trends in Mortality in a Large Regionalized Critical Care Transport Service.

OBJECTIVE: The epidemiology accompanying helicopter emergency medical services (HEMS) transport has evolved as agencies have matured and become integrated into regionalized health systems, as evidenced primarily by nationwide systems in Europe. System-level congruence between Europe and the United States, where HEMS is geographically fragmentary, is unclear. In this study, we provide a temporal, epidemiologic characterization of the largest standardized private, nonprofit HEMS system in the United States, STAT MedEvac. METHODS: We obtained comprehensive timing, procedure, and vital signs data from STAT MedEvac prehospital electronic patient care records for all adult patients transported to UPMC Health System hospitals in the period of January 2012 through October 2021. We linked these data with hospital electronic health records available through June 2018 to establish length of stay and vital status at discharge. RESULTS: We studied 90,960 transports and matched 62.8% (n = 57,128) to the electronic health record. The average patient age was 58.6 years ( 19 years), and most were male (57.9%). The majority of cases were interfacility transports (77.6%), and the most common general medical category was nontrauma (72.7%). Sixty-one percent of all patients received a prehospital intervention. Overall, hospital mortality was 15%, and the average hospital length of stay (LOS) was 8.8 days ( 10.0 days). Observed trends over time included increases in nontrauma transports, level of severity, and in-hospital mortality. In multivariable models, case severity and medical category correlated with the outcomes of mortality and LOS. CONCLUSION: In the largest standardized nonprofit HEMS system in the United States, patient mortality and hospital LOS increased over time, whereas the proportion of trauma patients and scene runs decreased.

Heart-Lung Interactions.

The pulmonary and cardiovascular systems have profound effects on each other. Overall cardiac function is determined by heart rate, preload, contractility, and afterload. Changes in lung volume, intrathoracic pressure (ITP), and hypoxemia can simultaneously change all of these four hemodynamic determinants for both ventricles and can even lead to cardiovascular collapse. Intubation using sedation depresses vasomotor tone. Also, the interdependence between right and left ventricles can be affected by lung volume-induced changes in pulmonary vascular resistance and the rise in ITP. An increase in venous return due to negative ITP during spontaneous inspiration can shift the septum to the left and cause a decrease in left ventricle compliance. During positive pressure ventilation, the increase in ITP causes a decrease in venous return (preload), minimizing ventricular interdependence and will decrease left ventricle afterload augmenting cardiac output. Thus, positive pressure ventilation is beneficial in acute heart failure patients and detrimental in hypovolemic patients where it can cause a significant decrease in venous return and cardiac output. Recently, this phenomenon has been used to assess patient's volume responsiveness to fluid by measuring pulse pressure variation and stroke volume variation. Heart-lung interaction is very dynamic and changes in lung volume, ITP, and oxygen level can have various effects on the cardiovascular system depending on preexisting cardiovascular function and volume status. Heart failure and either hypo or hypervolemia predispose to greater effects of ventilation of cardiovascular function and gas exchange. This review is an overview of the basics of heart-lung interaction.

Forecasting imminent atrial fibrillation in long-term electrocardiogram recordings.

BACKGROUND: Despite the morbidity associated with acute atrial fibrillation (AF), no models currently exist to forecast its imminent onset. We sought to evaluate the ability of deep learning to forecast the imminent onset of AF with sufficient lead time, which has important implications for inpatient care. METHODS: We utilized the Physiobank Long-Term AF Database, which contains 24-h, labeled ECG recordings from patients with a history of AF. AF episodes were defined as ≥5 min of sustained AF. Three deep learning models incorporating convolutional and transformer layers were created for forecasting, with two models focusing on the predictive nature of sinus rhythm segments and AF epochs separately preceding an AF episode, and one model utilizing all preceding waveform as input. Cross-validated performance was evaluated using area under time-dependent receiver operating characteristic curves (AUC(t)) at 7.5-, 15-, 30-, and 60-min lead times, precision-recall curves, and imminent AF risk trajectories. RESULTS: There were 367 AF episodes from 84 ECG recordings. All models showed average risk trajectory divergence of those with an AF episode from those without ∼15 min before the episode. Highest AUC was associated with the sinus rhythm model [AUC = 0.74; 7.5-min lead time], though the model using all preceding waveform data had similar performance and higher AUCs at longer lead times. CONCLUSIONS: In this proof-of-concept study, we demonstrated the potential utility of neural networks to forecast the onset of AF in long-term ECG recordings with a clinically relevant lead time. External validation in larger cohorts is required before deploying these models clinically.

Artificial Intelligence in Critical Care Medicine.

This article is one of ten reviews selected from the Annual Update in Intensive Care and Emergency Medicine 2022. Other selected articles can be found online at https://www.biomedcentral.com/collections/annualupdate2022 . Further information about the Annual Update in Intensive Care and Emergency Medicine is available from https://link.springer.com/bookseries/8901 .

A plea for personalization of the hemodynamic management of septic shock.

Although guidelines provide excellent expert guidance for managing patients with septic shock, they leave room for personalization according to patients' condition. Hemodynamic monitoring depends on the evolution phase: salvage, optimization, stabilization, and de-escalation. Initially during the salvage phase, monitoring to identify shock etiology and severity should include arterial pressure and lactate measurements together with clinical examination, particularly skin mottling and capillary refill time. Low diastolic blood pressure may trigger vasopressor initiation. At this stage, echocardiography may be useful to identify significant cardiac dysfunction. During the optimization phase, echocardiographic monitoring should be pursued and completed by the assessment of tissue perfusion through central or mixed-venous oxygen saturation, lactate, and carbon dioxide veno-arterial gradient. Transpulmonary thermodilution and the pulmonary artery catheter should be considered in the most severe patients. Fluid therapy also depends on shock phases. While administered liberally during the resuscitation phase, fluid responsiveness should be assessed during the optimization phase. During stabilization, fluid infusion should be minimized. In the de-escalation phase, safe fluid withdrawal could be achieved by ensuring tissue perfusion is preserved. Norepinephrine is recommended as first-line vasopressor therapy, while vasopressin may be preferred in some patients. Essential questions remain regarding optimal vasopressor selection, combination therapy, and the most effective and safest escalation. Serum renin and the angiotensin I/II ratio may identify patients who benefit most from angiotensin II. The optimal therapeutic strategy for shock requiring high-dose vasopressors is scant. In all cases, vasopressor therapy should be individualized, based on clinical evaluation and blood flow measurements to avoid excessive vasoconstriction. Inotropes should be considered in patients with decreased cardiac contractility associated with impaired tissue perfusion. Based on pharmacologic properties, we suggest as the first test a limited dose of dobutamine, to add enoximone or milrinone in the second line and substitute or add levosimendan if inefficient. Regarding adjunctive therapies, while hydrocortisone is nowadays advised in patients receiving high doses of vasopressors, patients responding to corticosteroids may be identified in the future by the analysis of selected cytokines or specific transcriptomic endotypes. To conclude, although some general rules apply for shock management, a personalized approach should be considered for hemodynamic monitoring and support.

Effective hemodynamic monitoring.

Hemodynamic monitoring is the centerpiece of patient monitoring in acute care settings. Its effectiveness in terms of improved patient outcomes is difficult to quantify. This review focused on effectiveness of monitoring-linked resuscitation strategies from: (1) process-specific monitoring that allows for non-specific prevention of new onset cardiovascular insufficiency (CVI) in perioperative care. Such goal-directed therapy is associated with decreased perioperative complications and length of stay in high-risk surgery patients. (2) Patient-specific personalized resuscitation approaches for CVI. These approaches including dynamic measures to define volume responsiveness and vasomotor tone, limiting less fluid administration and vasopressor duration, reduced length of care. (3) Hemodynamic monitoring to predict future CVI using machine learning approaches. These approaches presently focus on predicting hypotension. Future clinical trials assessing hemodynamic monitoring need to focus on process-specific monitoring based on modifying therapeutic interventions known to improve patient-centered outcomes.

Intelligent Clinical Decision Support.

Early recognition of pathologic cardiorespiratory stress and forecasting cardiorespiratory decompensation in the critically ill is difficult even in highly monitored patients in the Intensive Care Unit (ICU). Instability can be intuitively defined as the overt manifestation of the failure of the host to adequately respond to cardiorespiratory stress. The enormous volume of patient data available in ICU environments, both of high-frequency numeric and waveform data accessible from bedside monitors, plus Electronic Health Record (EHR) data, presents a platform ripe for Artificial Intelligence (AI) approaches for the detection and forecasting of instability, and data-driven intelligent clinical decision support (CDS). Building unbiased, reliable, and usable AI-based systems across health care sites is rapidly becoming a high priority, specifically as these systems relate to diagnostics, forecasting, and bedside clinical decision support. The ICU environment is particularly well-positioned to demonstrate the value of AI in saving lives. The goal is to create AI models embedded in a real-time CDS for forecasting and mitigation of critical instability in ICU patients of sufficient readiness to be deployed at the bedside. Such a system must leverage multi-source patient data, machine learning, systems engineering, and human action expertise, the latter being key to successful CDS implementation in the clinical workflow and evaluation of bias. We present one approach to create an operationally relevant AI-based forecasting CDS system.

Automated Assessment of Cardiovascular Sufficiency Using Non-Invasive Physiological Data.

For fluid resuscitation of critically ill individuals to be effective, it must be well calibrated in terms of timing and dosages of treatments. In current practice, the cardiovascular sufficiency of patients during fluid resuscitation is determined using primarily invasively measured vital signs, including Arterial Pressure and Mixed Venous Oxygen Saturation (SvO2), which may not be available in outside-of-hospital settings, particularly in the field when treating subjects injured in traffic accidents or wounded in combat where only non-invasive monitoring is available to drive care. In this paper, we propose (1) a Machine Learning (ML) approach to estimate the sufficiency utilizing features extracted from non-invasive vital signs and (2) a novel framework to address the detrimental impact of inter-patient diversity on the ability of ML models to generalize well to unseen subjects. Through comprehensive evaluation on the physiological data collected in laboratory animal experiments, we demonstrate that the proposed approaches can achieve competitive performance on new patients using only non-invasive measurements. These characteristics enable effective monitoring of fluid resuscitation in real-world acute settings with limited monitoring resources and can help facilitate broader adoption of ML in this important subfield of healthcare.

Accuracy of identifying hospital acquired venous thromboembolism by administrative coding: implications for big data and machine learning research.

Big data analytics research using heterogeneous electronic health record (EHR) data requires accurate identification of disease phenotype cases and controls. Overreliance on ground truth determination based on administrative data can lead to biased and inaccurate findings. Hospital-acquired venous thromboembolism (HA-VTE) is challenging to identify due to its temporal evolution and variable EHR documentation. To establish ground truth for machine learning modeling, we compared accuracy of HA-VTE diagnoses made by administrative coding to manual review of gold standard diagnostic test results. We performed retrospective analysis of EHR data on 3680 adult stepdown unit patients identifying HA-VTE. International Classification of Diseases, Ninth Revision (ICD-9-CM) codes for VTE were identified. 4544 radiology reports associated with VTE diagnostic tests were screened using terminology extraction and then manually reviewed by a clinical expert to confirm diagnosis. Of 415 cases with ICD-9-CM codes for VTE, 219 were identified with acute onset type codes. Test report review identified 158 new-onset HA-VTE cases. Only 40% of ICD-9-CM coded cases (n = 87) were confirmed by a positive diagnostic test report, leaving the majority of administratively coded cases unsubstantiated by confirmatory diagnostic test. Additionally, 45% of diagnostic test confirmed HA-VTE cases lacked corresponding ICD codes. ICD-9-CM coding missed diagnostic test-confirmed HA-VTE cases and inaccurately assigned cases without confirmed VTE, suggesting dependence on administrative coding leads to inaccurate HA-VTE phenotyping. Alternative methods to develop more sensitive and specific VTE phenotype solutions portable across EHR vendor data are needed to support case-finding in big-data analytics.

Activation of AMP-activated protein kinase during sepsis/inflammation improves survival by preserving cellular metabolic fitness.

The purpose was to determine the role of AMPK activation in the renal metabolic response to sepsis, the development of sepsis-induced acute kidney injury (AKI) and on survival. In a prospective experimental study, 167 10- to 12-week-old C57BL/6 mice underwent cecal ligation and puncture (CLP) and human proximal tubule epithelial cells (TEC; HK2) were exposed to inflammatory mix (IM), a combination of lipopolysaccharide (LPS) and high mobility group box 1 (HMGB1). Renal/TEC metabolic fitness was assessed by monitoring the expression of drivers of oxidative phosphorylation (OXPHOS), the rates of utilization of OXPHOS/glycolysis in response to metabolic stress, and mitochondrial function by measuring O2 consumption rates (OCR) and the membrane potential (Δψm ). Sepsis/IM resulted in AKI, increased mortality, and in renal AMPK activation 6-24 hours after CLP/IM. Pharmacologic activation of AMPK with 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) or metformin during sepsis improved the survival, while AMPK inhibition with Compound C increased mortality, impaired mitochondrial respiration, decreased OCR, and disrupted TEC metabolic fitness. AMPK-driven protection was associated with increased Sirt 3 expression and restoration of metabolic fitness. Renal AMPK activation in response to sepsis/IM is an adaptive mechanism that protects TEC, organs, and the host by preserving mitochondrial function and metabolic fitness likely through Sirt3 signaling.

Hemodynamic monitoring in cardiogenic shock.

PURPOSE OF REVIEW: Cardiogenic shock remains a major cause of mortality today. With recent advancements in invasive mechanical support strategies, reperfusion practice, and a new classification scheme is proposed for cardiogenic shock, an updated review of the latest hemodynamic monitoring techniques is important. RECENT FINDINGS: Multiple recent studies have emerged supporting the use of pulmonary artery catheters in the cardiogenic shock population. Data likewise continues to emerge on the use of echocardiography and biomarker measurement in the care of these patients. SUMMARY: The integration of multiple forms of hemodynamic monitoring, spanning noninvasive and invasive modalities, is important in the diagnosis, staging, initial treatment, and subsequent management of the cardiogenic shock patient.

Dynamic right and left ventricular interactions in the pig.

NEW FINDINGS: What is the central question of this study? Are the mechanisms that cause ventricular interdependence different when due to primary right to left ventricular pressure loading? What is the main finding and its importance? An instantaneous selective increase in aortic pressure causes an immediate increase in right ventricular end-systolic pressure independent of the pericardium, whereas a selective increase in pulmonary artery pressure decreases left ventricular diastolic compliance owing to a subsequent increasing right ventricular end-diastolic volume as a function of an intact pericardium limiting biventricular volume. Changes in contraction synchrony of either ventricle do not appear to be causing these effects. ABSTRACT: I characterized the dynamic factors determining ventricular interdependence with and without the pericardium. I measured right (RV) and left ventricular (LV) pressures and volumes simultaneously using conductance catheters in seven pentobarbitone-anaesthetized open-chested 5- to 7-week-old piglets. I studied these effects during apnoea, inferior vena caval occlusion and rapid partial aortic and pulmonary arterial occlusions. Conductance catheter-defined long-axis regional volumes were assessed to define regional contractile synchrony. Closed-pericardium measures were made from an initial (baseline) volume, then after two 20 ml kg-1 fluid loads followed by an open-pericardium step. Baseline RV and LV volumes were similar. Aortic occlusion increased LV pressures and volumes and RV end-systolic pressure such that RV end-systolic elastance increased without changes in RV contraction synchrony, not affected by the pericardium. Pulmonary artery occlusion increased RV end-systolic pressure but not end-systolic volume. On the subsequent beat, RV end-diastolic pressure increased, whereas LV end-diastolic volume and diastolic compliance decreased. These effects were attenuated by opening the pericardium. Contraction synchrony across longitudinal segments was unaltered by either aortic or pulmonary artery occlusion. I conclude that the determinants of systolic and diastolic ventricular interdependence are different. Increasing RV pressures causes diastolic RV-to-LV interdependence, decreasing LV diastolic compliance and dependent on an intact pericardium. An increase in LV end-systolic pressure increases RV end-systolic elastance independent of the pericardium and has a minimal effect on RV diastolic function or contraction synchrony.

How to assess ventriculoarterial coupling in sepsis.

PURPOSE OF REVIEW: We will highlight the role of ventriculoarterial coupling in the pathophysiology of sepsis and how to assess it. RECENT FINDINGS: Most septic patients show a ventriculoarterial uncoupling at the time of diagnosis with arterial elastance (Ea) greater than left ventricle (LV) end-systolic elastance (Ees), often despite arterial hypotension. Ventriculoarterial coupling levels predict the cardiovascular response to resuscitation in this heterogeneously responding population. SUMMARY: Ventriculoarterial coupling is quantified as the ratio of Ea to Ees. The efficiency of the cardiovascular function is optimal when Ea/Ees is near one. When the hydraulic load of the arterial system is excessive either from increased vasomotor tone, decreased LV contractility or both, Ea/Ees becomes greater than 1 (i.e. ventriculoarterial decoupling), and cardiac efficiency decreases leading to heart failure, loss of volume responsiveness, and if sustained, increased mortality. Noninvasive echocardiographic techniques when linked with arterial pressure monitoring allow for the bedside estimates of both Ea and Ees. Studies using this approach have documented the key role ventriculoarterial coupling has defining initial cardiovascular state, response to therapy and outcome from critical illness. Sequential monitoring of ventriculoarterial coupling at the bedside offers a unique opportunity to assess relevant cardiovascular determinants in septic patients requiring resuscitation.

Heterogeneity of Cardiovascular Response to Standardized Sepsis Resuscitation.

This article is one of ten reviews selected from the Annual Update in Intensive Care and Emergency Medicine 2020. Other selected articles can be found online at https://www.biomedcentral.com/collections/annualupdate2020. Further information about the Annual Update in Intensive Care and Emergency Medicine is available from http://www.springer.com/series/8901.

Prediction of hypotension events with physiologic vital sign signatures in the intensive care unit.

BACKGROUND: Even brief hypotension is associated with increased morbidity and mortality. We developed a machine learning model to predict the initial hypotension event among intensive care unit (ICU) patients and designed an alert system for bedside implementation. MATERIALS AND METHODS: From the Medical Information Mart for Intensive Care III (MIMIC-3) dataset, minute-by-minute vital signs were extracted. A hypotension event was defined as at least five measurements within a 10-min period of systolic blood pressure ≤ 90 mmHg and mean arterial pressure ≤ 60 mmHg. Using time series data from 30-min overlapping time windows, a random forest (RF) classifier was used to predict risk of hypotension every minute. Chronologically, the first half of extracted data was used to train the model, and the second half was used to validate the trained model. The model's performance was measured with area under the receiver operating characteristic curve (AUROC) and area under the precision recall curve (AUPRC). Hypotension alerts were generated using risk score time series, a stacked RF model. A lockout time were applied for real-life implementation. RESULTS: We identified 1307 subjects (1580 ICU stays) as the hypotension group and 1619 subjects (2279 ICU stays) as the non-hypotension group. The RF model showed AUROC of 0.93 and 0.88 at 15 and 60 min, respectively, before hypotension, and AUPRC of 0.77 at 60 min before. Risk score trajectories revealed 80% and > 60% of hypotension predicted at 15 and 60 min before the hypotension, respectively. The stacked model with 15-min lockout produced on average 0.79 alerts/subject/hour (sensitivity 92.4%). CONCLUSION: Clinically significant hypotension events in the ICU can be predicted at least 1 h before the initial hypotension episode. With a highly sensitive and reliable practical alert system, a vast majority of future hypotension could be captured, suggesting potential real-life utility.

Parsimony of Hemodynamic Monitoring Data Sufficient for the Detection of Hemorrhage.

BACKGROUND: Individualized hemodynamic monitoring approaches are not well validated. Thus, we evaluated the discriminative performance improvement that might occur when moving from noninvasive monitoring (NIM) to invasive monitoring and with increasing levels of featurization associated with increasing sampling frequency and referencing to a stable baseline to identify bleeding during surgery in a porcine model. METHODS: We collected physiologic waveform (WF) data (250 Hz) from NIM, central venous (CVC), arterial (ART), and pulmonary arterial (PAC) catheters, plus mixed venous O2 saturation and cardiac output from 38 anesthetized Yorkshire pigs bled at 20 mL/min until a mean arterial pressure of 30 mm Hg following a 30-minute baseline period. Prebleed physiologic data defined a personal stable baseline for each subject independently. Nested models were evaluated using simple hemodynamic metrics (SM) averaged over 20-second windows and sampled every minute, beat to beat (B2B), and WF using Random Forest Classification models to identify bleeding with or without normalization to personal stable baseline, using a leave-one-pig-out cross-validation to minimize model overfitting. Model hyperparameters were tuned to detect stable or bleeding states. Bleeding models were compared use both each subject's personal baseline and a grouped-average (universal) baseline. Timeliness of bleed onset detection was evaluated by comparing the tradeoff between a low false-positive rate (FPR) and shortest time to bleed detection. Predictive performance was evaluated using a variant of the receiver operating characteristic focusing on minimizing FPR and false-negative rates (FNR) for true-positive and true-negative rates, respectively. RESULTS: In general, referencing models to a personal baseline resulted in better bleed detection performance for all catheters than using universal baselined data. Increasing granularity from SM to B2B and WF progressively improved bleeding detection. All invasive monitoring outperformed NIM for both time to bleeding detection and low FPR and FNR. In that regard, when referenced to personal baseline with SM analysis, PAC and ART + PAC performed best; for B2B CVC, PAC and ART + PAC performed best; and for WF PAC, CVC, ART + CVC, and ART + PAC performed equally well and better than other monitoring approaches. Without personal baseline, NIM performed poorly at all levels, while all catheters performed similarly for SM, with B2B PAC and ART + PAC performing the best, and for WF PAC, ART, ART + CVC, and ART + PAC performed equally well and better than the other monitoring approaches. CONCLUSIONS: Increasing hemodynamic monitoring featurization by increasing sampling frequency and referencing to personal baseline markedly improves the ability of invasive monitoring to detect bleed.

Estimating Surgical Blood Loss Volume Using Continuously Monitored Vital Signs.

Background: There are currently no effective and accurate blood loss volume (BLV) estimation methods that can be implemented in operating rooms. To improve the accuracy and reliability of BLV estimation and facilitate clinical implementation, we propose a novel estimation method using continuously monitored photoplethysmography (PPG) and invasive arterial blood pressure (ABP). Methods: Forty anesthetized York Pigs (31.82 ± 3.52 kg) underwent a controlled hemorrhage at 20 mL/min until shock development was included. Machine-learning-based BLV estimation models were proposed and tested on normalized features derived by vital signs. Results: The results showed that the mean ± standard deviation (SD) for estimating BLV against the reference BLV of our proposed random-forest-derived BLV estimation models using PPG and ABP features, as well as the combination of ABP and PPG features, were 11.9 ± 156.2, 6.5 ± 161.5, and 7.0 ± 139.4 mL, respectively. Compared with traditional hematocrit computation formulas (estimation error: 102.1 ± 313.5 mL), our proposed models outperformed by nearly 200 mL in SD. Conclusion: This is the first attempt at predicting quantitative BLV from noninvasive measurements. Normalized PPG features are superior to ABP in accurately estimating early-stage BLV, and normalized invasive ABP features could enhance model performance in the event of a massive BLV.