Improving the precision of shock resuscitation by predicting fluid responsiveness with machine learning and arterial blood pressure waveform data.

Fluid bolus therapy (FBT) is fundamental to the management of circulatory shock in critical care but balancing the benefits and toxicities of FBT has proven challenging in individual patients. Improved predictors of the hemodynamic response to a fluid bolus, commonly referred to as a fluid challenge, are needed to limit non-beneficial fluid administration and to enable automated clinical decision support and patient-specific precision critical care management. In this study we retrospectively analyzed data from 394 fluid boluses from 58 pigs subjected to either hemorrhagic or distributive shock. All animals had continuous blood pressure and cardiac output monitored throughout the study. Using this data, we developed a machine learning (ML) model to predict the hemodynamic response to a fluid challenge using only arterial blood pressure waveform data as the input. A Random Forest binary classifier referred to as the ML fluid responsiveness algorithm (MLFRA) was trained to detect fluid responsiveness (FR), defined as a ≥ 15% change in cardiac stroke volume after a fluid challenge. We then compared its performance to pulse pressure variation, a commonly used metric of FR. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC), confusion matrix metrics, and calibration curves plotting predicted probabilities against observed outcomes. Across multiple train/test splits and feature selection methods designed to assess performance in the setting of small sample size conditions typical of large animal experiments, the MLFRA achieved an average AUROC, recall (sensitivity), specificity, and precision of 0.82, 0.86, 0.62. and 0.76, respectively. In the same datasets, pulse pressure variation had an AUROC, recall, specificity, and precision of 0.73, 0.91, 0.49, and 0.71, respectively. The MLFRA was generally well-calibrated across its range of predicted probabilities and appeared to perform equally well across physiologic conditions. These results suggest that ML, using only inputs from arterial blood pressure monitoring, may substantially improve the accuracy of predicting FR compared to the use of pulse pressure variation. If generalizable, these methods may enable more effective, automated precision management of critically ill patients with circulatory shock.

Precision Automated Critical Care Management: Closed-loop critical care for the treatment of distributive shock in a swine model of ischemia-reperfusion.

BACKGROUND: Goal-directed blood pressure management in the intensive care unit can improve trauma outcomes but is labor-intensive. Automated critical care systems can deliver scaled interventions to avoid excessive fluid or vasopressor administration. We compared a first-generation automated drug and fluid delivery platform, Precision Automated Critical Care Management (PACC-MAN), to a more refined algorithm, incorporating additional physiologic inputs and therapeutics. We hypothesized that the enhanced algorithm would achieve equivalent resuscitation endpoints with less crystalloid utilization in the setting of distributive shock. METHODS: Twelve swine underwent 30% hemorrhage and 30 minutes of aortic occlusion to induce an ischemia-reperfusion injury and distributive shock state. Next, animals were transfused to euvolemia and randomized into a standardized critical care (SCC) of PACC-MAN or an enhanced version (SCC+) for 4.25 hours. SCC+ incorporated lactate and urine output to assess global response to resuscitation and added vasopressin as an adjunct to norepinephrine at certain thresholds. Primary and secondary outcomes were decreased crystalloid administration and time at goal blood pressure, respectively. RESULTS: Weight-based fluid bolus volume was lower in SCC+ compared with SCC (26.9 mL/kg vs. 67.5 mL/kg, p = 0.02). Cumulative norepinephrine dose required was not significantly different (SCC+: 26.9 µg/kg vs. SCC: 13.76 µg/kg, p = 0.24). Three of 6 animals (50%) in SCC+ triggered vasopressin as an adjunct. Percent time spent between 60 mm Hg and 70 mm Hg, terminal creatinine and lactate, and weight-adjusted cumulative urine output were equivalent. CONCLUSION: Refinement of the PACC-MAN algorithm decreased crystalloid administration without sacrificing time in normotension, reducing urine output, increasing vasopressor support, or elevating biomarkers of organ damage. Iterative improvements in automated critical care systems to achieve target hemodynamics in a distributive-shock model are feasible.

Automated partial resuscitative endovascular balloon occlusion of the aorta reduces blood loss and hypotension in a highly lethal porcine liver injury model.

BACKGROUND: Partial and intermittent resuscitative endovascular balloon occlusion of the aorta (pREBOA and iREBOA, respectively) are lifesaving techniques designed to extend therapeutic duration, mitigate ischemia, and bridge patients to definitive hemorrhage control. We hypothesized that automated pREBOA balloon titration compared with automated iREBOA would reduce blood loss and hypotensive episodes over a 90-minute intervention phase compared with iREBOA in an uncontrolled liver hemorrhage swine model. METHODS: Twenty-four pigs underwent an uncontrolled hemorrhage by liver transection and were randomized to automated pREBOA (n = 8), iREBOA (n = 8), or control (n = 8). Once hemorrhagic shock criteria were met, controls had the REBOA catheter removed and received transfusions only for hypotension. The REBOA groups received 90 minutes of either iREBOA or pREBOA therapy. Surgical hemostasis was obtained, hemorrhage volume was quantified, and animals were transfused to euvolemia and then underwent 1.5 hours of automated critical care. RESULTS: The control group had significantly higher mortality rate (5 of 8) compared with no deaths in both REBOA groups, demonstrating that the liver injury is highly lethal ( p = 0.03). During the intervention phase, animals in the iREBOA group spent a greater proportion of time in hypotension than the pREBOA group (20.7% [16.2-24.8%] vs. 0.76% [0.43-1.14%]; p < 0.001). The iREBOA group required significantly more transfusions than pREBOA (21.0 [20.0-24.9] mL/kg vs. 12.1 [9.5-13.9] mL/kg; p = 0.01). At surgical hemostasis, iREBOA had significantly higher hemorrhage volumes compared with pREBOA (39.2 [29.7-44.95] mL/kg vs. 24.7 [21.6-30.8] mL/kg; p = 0.04). CONCLUSION: Partial REBOA animals spent significantly less time at hypotension and had decreased transfusions and blood loss. Both pREBOA and iREBOA prevented immediate death compared with controls. Further refinement of automated pREBOA is necessary, and controller algorithms may serve as vital control inputs for automated transfusion. LEVEL OF EVIDENCE: Therapeutic/Care Management; Level III.

Hardware and software implementation of POCT1-A for integration of point of care testing in research.

Point of care testing (POCT) is increasingly utilized in clinical medicine. Small, portable testing devices can now deliver reliable and accurate diagnostic results during a patient encounter. With these increases in POCT, the issue of data and results management quickly emerges. Results need to be cataloged accurately and efficiently while the providers/support staff are simultaneously managing patient encounters. The integration of electronic medical records (EMR) as data repositories requires that point of care testing data imports automatically into the EMR. POCT1-A was developed as a standard communication language for POCT device manufacturers to streamline automatic data import integration. While all modern POCT devices are built with this connectivity, the systems that provide the integration layer are often proprietary and require a fee for service. In the research environment, there is not enough throughput to justify the practical investment in these data management architectures. Moreover, researcher needs are different and unique compared to data management systems for clinicians. To meet this need, we developed a novel hardware and software connectivity solution using commercially available components to automate data management from a point-of-care blood biochemical analyzer during a critical care study in the preclinical research environment.

Near-Infrared Spectroscopy for Determination of Cardiac Output Augmentation in a Swine Model of Ischemia-Reperfusion Injury.

CONTEXT: Near infrared spectroscopy (NIRS) is a noninvasive tool for assessing local oxygen balance. In circulatory shock, the microcirculatory environment as measured by NIRS during resuscitation may provide additional diagnostic tools of value to the critical care physician. HYPOTHESIS: To assess whether a relative increase in peripheral NIRS was correlated with a clinically relevant increase in cardiac output following a fluid bolus in a swine model of shock. METHODS AND MODELS: Nine healthy young adult swine with median weight 80 kg (interquartile range, 75-83 kg) were anesthetized and surgically instrumented. They underwent a controlled hemorrhage of 20% of their blood volume followed by partial or complete aortic occlusion to create a variable ischemia-reperfusion injury. Next, the animals underwent four 500-mL plasmalyte boluses over 9 minutes each followed by a 6-minute pause. The animal then underwent a 25% mixed auto/homologous blood transfusion followed by four more 500 mL plasmalyte boluses over 9 minutes. Finally, the animals underwent a 25% mixed auto/homologous blood transfusion followed by an additional four rounds of 500-mL plasmalyte boluses over 9 minutes. Left thoracic limb NIRS, descending thoracic aortic flow (dAF), arterial blood pressure (MAP), central venous pressure (CVP), and mixed central venous oxygen saturation (Svo2) were measured continuously for comparison. RESULTS: The area under the receiver operating curve for an increase in dAF of 10% in response to a 500 mL bolus based on a percent increase in the proximal NIRS was 0.82 with 95% CI, 0.72-0.91; Svo2, 0.86 with 95% CI, 0.78-0.95; MAP, 0.75 with 95% CI, 0.65-0.85 and CVP, 0.64 with 95% CI, 0.53-0.76. INTERPRETATION AND CONCLUSIONS: A dynamic relative increase in NIRS in response to a crystalloid challenge has moderate discriminatory power for cardiac output augmentation during shock in a swine model of ischemia-reperfusion injury. NIRS performed as well as invasive measurements (Svo2 and MAP) and better than CVP.

Endovascular Perfusion Augmentation After Resuscitative Endovascular Balloon Occlusion of the Aorta Improves Renal Perfusion and Decreases Vasopressors.

INTRODUCTION: Resuscitative endovascular balloon occlusion of the aorta (REBOA) causes a severe ischemia-reperfusion injury. Endovascular Perfusion Augmentation for Critical Care (EPACC) has emerged as a hemodynamic/mechanical adjunct to vasopressors and crystalloid for the treatment of post-REBOA ischemia-reperfusion injury. The objective of the study is to examine the impact of EPACC as a tool for a wean from complete REBOA compared to standard resuscitation techniques. METHODS: Nine swine underwent anesthesia and then a controlled 30% blood volume hemorrhage with 30 min of supraceliac total aortic occlusion to create an ischemia-reperfusion injury. Animals were randomized to standardized critical care (SCC) or 90 min of EPACC followed by SCC. The critical care phase lasted 270 min after injury. Hemodynamic markers and laboratory values of ischemia were recorded. RESULTS: During the first 90 min the intervention phase SCC spent 60% (54%-73%) and EPACC spent 91% (88%-92%) of the time avoiding proximal hypotension (<60 mm Hg), P = 0.03. There was also a statistically significant decrease in cumulative norepinephrine dose at the end of the experiment between SCC (80.89 mcg/kg) versus EPACC (22.03 mcg/kg), P = 0.03. Renal artery flow during EPACC was similar compared to SCC during EPACC, P = 0.19. But during the last hour of the experiment (after removal of aortic balloon) the renal artery flow in EPACC (2.9 mL/kg/min) was statistically significantly increased compared to SCC (1.57 mL/min/kg), P = 0.03. There was a statistically significant decrease in terminal creatinine in the EPACC (1.7 mg/dL) compared to SCC (2.1 mg/dL), P = 0.03. CONCLUSIONS: The 90 min of EPACC as a weaning adjunct in the setting of a severe ischemia-reperfusion injury after complete supraceliac REBOA provides improved renal flow with improvement in terminal creatinine compared to SCC with stabilized proximal hemodynamics and decreased vasopressor dose.

Closed-loop automated critical care as proof-of-concept study for resuscitation in a swine model of ischemia-reperfusion injury.

BACKGROUND: Volume expansion and vasopressors for the treatment of shock is an intensive process that requires frequent assessments and adjustments. Strict blood pressure goals in multiple physiologic states of shock (traumatic brain injury, sepsis, and hemorrhagic) have been associated with improved outcomes. The availability of continuous physiologic data is amenable to closed-loop automated critical care to improve goal-directed resuscitation. METHODS: Five adult swine were anesthetized and subjected to a controlled 30% estimated total blood volume hemorrhage followed by 30 min of complete supra-celiac aortic occlusion and then autotransfusion back to euvolemia with removal of aortic balloon. The animals underwent closed-loop critical care for 255 min after removal of the endovascular aortic balloon. The closed-loop critical care algorithm used proximal aortic pressure and central venous pressure as physiologic input data. The algorithm had the option to provide programmatic control of pumps for titration of vasopressors and weight-based crystalloid boluses (5 ml/kg) to maintain a mean arterial pressure between 60 and 70 mmHg. RESULTS: During the 255 min of critical care the animals experienced hypotension (< 60 mmHg) 15.3% (interquartile range: 8.6-16.9%), hypertension (> 70 mmHg) 7.7% (interquartile range: 6.7-9.4%), and normotension (60-70 mmHg) 76.9% (interquartile range: 76.5-81.2%) of the time. Excluding the first 60 min of the critical care phase the animals experienced hypotension 1.0% (interquartile range: 0.5-6.7%) of the time. Median intervention rate was 8.47 interventions per hour (interquartile range: 7.8-9.2 interventions per hour). The proportion of interventions was 61.5% (interquartile range: 61.1-66.7%) weight-based crystalloid boluses and 38.5% (interquartile range: 33.3-38.9%) titration of vasopressors. CONCLUSION: This autonomous critical care platform uses critical care adjuncts in an ischemia-reperfusion injury model, utilizing goal-directed closed-loop critical care algorithm and device actuation. This description highlights the potential for this approach to deliver nuanced critical care in the ICU environment, thereby optimizing resuscitative efforts and expanding capabilities through cognitive offloading. Future efforts will focus on optimizing this platform through comparative studies of inputs, therapies, and comparison to manual critical care.

Endovascular Perfusion Augmentation for Critical Care Decreases Vasopressor Requirements while Maintaining Renal Perfusion.

BACKGROUND: Ischemia reperfusion injury causes a profound hyperdynamic distributive shock. Endovascular perfusion augmentation for critical care (EPACC) has emerged as a hemodynamic adjunct to vasopressors and crystalloid. The objective of this study was to examine varying levels of mechanical support for the treatment of ischemiareperfusion injury in swine. METHODS: Fifteen swine underwent anesthesia and then a controlled 30% blood volume hemorrhage followed by 30 min of supra-celiac aortic occlusion to create an ischemia-reperfusion injury Animals were randomized to standardized critical care (SCC), EPACC with low threshold (EPACC-Low), and EPACC with high threshold (EPACC-High). The intervention phase lasted 270 min after injury Hemodynamic markers and laboratory values of ischemia were recorded. RESULTS: During the intervention phase, SCC spent 82.4% of the time avoiding proximal hypotension (>60 mm Hg), while EPACC-Low spent 97.6% and EPACC-High spent 99.5% of the time avoiding proximal hypotension, P  < 0.001. Renal artery flow was statistically increased in EPACC-Low compared with SCC (2.29 mL/min/kg vs. 1.77 mL/ min/kg, P  < 0.001), while renal flow for EPACC-High was statistically decreased compared with SCC (1.25 mL/min/kg vs. 1.77 mL/min/kg, P  < 0.001). EPACC animals required less intravenous norepinephrine, (EPACC-Low: 16.23mcg/kg and EPACC-High: 13.72 mcg/kg), compared with SCC (59.45 mcg/kg), P = 0.049 and P = 0.013 respectively. CONCLUSIONS: Compared with SCC, EPACC-High and EPACC-Low had decreased norepinephrine requirements with decreased frequency of proximal hypotension. EPACC-Low paradoxically had increased renal perfusion despite having a mechanical resistor in the aorta proximal to the renal arteries. This is the first description of low volume mechanical hemodynamic support in the setting of profound shock from ischemia-reperfusion injury in swine demonstrating stabilized proximal hemodynamics and augmented distal perfusion.

The physiology of aortic flow and pressures during partial resuscitative endovascular balloon occlusion of the aorta in a swine model of hemorrhagic shock.

BACKGROUND: Partial resuscitative endovascular balloon occlusion of the aorta (REBOA) has shown promise as a method to extend REBOA, but there lacks a standard definition of the technique. The purpose of this study was to investigate the relationships between distal and proximal mean arterial pressure (MAP) and distal aortic flow past a REBOA catheter. We hypothesize that a relationship between distal aortic flow and distal MAP in Zone 1 partial REBOA (pREBOA) is conserved and that there is no apparent relationship between aortic flow and proximal MAP. METHODS: A retrospective data analysis of swine was performed. Cohort 1 underwent 20% controlled hemorrhage and then randomized to aortic flow of 400 mL/min or complete occlusion for 20 minutes (n = 11). Cohort 2 underwent 30% controlled hemorrhage followed by complete aortic occlusion for 30 minutes (n = 29). Then, they all underwent REBOA wean in a similar stepwise fashion. Blood pressure was collected from above (proximal) and below (distal) the REBOA balloon. Aortic flow was measured using a surgically implanted supraceliac aortic perivascular flow probe. The time period of balloon wean was taken as the time point of interest. RESULTS: A linear relationship between distal MAP and aortic flow was observed ( R2 value, 0.80), while no apparent relationship appeared between proximal MAP and aortic flow ( R2 value, 0.29). The repeated-measures correlation coefficient for distal MAP (0.94; 95% confidence interval, 0.94-0.94) was greater than proximal MAP (-0.73; 95% confidence interval, -0.74 to -0.72). CONCLUSION: The relationship between MAP and flow will be a component of next-generation pREBOA control inputs. This study provides evidence that pREBOA techniques should rely on distal rather than proximal MAP for control of distal aortic flow. These data could inform future inquiry into optimal flow rates and parameters based on distal MAP in both translational and clinical contexts.