Steven L. Moulton

Use of advanced machine-learning techniques for noninvasive monitoring of hemorrhage.

BACKGROUND Hemorrhagic shock is a leading cause of death in both civilian and battlefield trauma. Currently available medical monitors provide measures of standard vital signs that are insensitive and nonspecific. More important, hypotension and other signs and symptoms of shock can appear when it may be too late to apply effective life-saving interventions. The resulting challenge is that early diagnosis is difficult because hemorrhagic shock is first recognized by late-responding vital signs and symptoms. The purpose of these experiments was to test the hypothesis that state-of-the-art machine-learning techniques, when integrated with novel non-invasive monitoring technologies, could detect early indicators of blood volume loss and impending circulatory failure in conscious, healthy humans who experience reduced central blood volume. METHODS Humans were exposed to progressive reductions in central blood volume using lower body negative pressure as a model of hemorrhage until the onset of hemodynamic decompensation. Continuous, noninvasively measured hemodynamic signals were used for the development of machine-learning algorithms. Accuracy estimates were obtained by building models using signals from all but one subject and testing on that subject. This process was repeated, each time using a different subject. RESULTS The model was 96.5% accurate in predicting the estimated amount of reduced central blood volume, and the correlation between predicted and actual lower body negative pressure level for hemodynamic decompensation was 0.89. CONCLUSIONS Machine modeling can accurately identify reduced central blood volume and predict impending hemodynamic decompensation (shock onset) in individuals. Such a capability can provide decision support for earlier intervention.

Running on empty? The compensatory reserve index.

BACKGROUND Hemorrhage is a leading cause of traumatic death. We hypothesized that state-of-the-art feature extraction and machine learning techniques could be used to discover, detect, and continuously trend beat-to-beat changes in arterial pulse waveforms associated with the progression to hemodynamic decompensation. METHODS We exposed 184 healthy humans to progressive central hypovolemia using lower-body negative pressure to the point of hemodynamic decompensation (systolic blood pressure > 80 mm Hg with or without bradycardia). Initial models were developed using continuous noninvasive blood pressure waveform data. The resulting algorithm calculates a compensatory reserve index (CRI), where 1 represents supine normovolemia and 0 represents the circulatory volume at which hemodynamic decompensation occurs (i.e., “running on empty”). Values between 1 and 0 indicate the proportion of reserve remaining before hemodynamic decompensation—much like the fuel gauge of a car indicates the amount of fuel remaining in the tank. A CRI estimate is produced after the first 30 heart beats, followed by a new CRI estimate after each subsequent beat. RESULTS The CRI model with a 30-beat window has an absolute difference between actual and expected time to decompensation of 0.1, with a SD of 0.09. The model distinguishes individuals with low tolerance to reduced central blood volume (i.e., those most likely to develop early shock) from those with high tolerance and are able to estimate how near or far an individual may be from hemodynamic decompensation. CONCLUSION Machine modeling can quickly and accurately detect and trend central blood volume reduction in real time during the compensatory phase of hemorrhage as well as estimate when an individual is “running on empty” and will decompensate (CRI, 0), well in advance of meaningful changes in traditional vital signs.

Individual-Specific, Beat-to-beat Trending of Significant Human Blood Loss: The Compensatory Reserve.

ABSTRACT Current monitoring technologies are unable to detect early, compensatory changes that are associated with significant blood loss. We previously introduced a novel algorithm to calculate the Compensatory Reserve Index (CRI) based on the analysis of arterial waveform features obtained from photoplethysmogram recordings. In the present study, we hypothesized that the CRI would provide greater sensitivity and specificity to detect blood loss compared with traditional vital signs and other hemodynamic measures. Continuous noninvasive vital sign waveform data, including CRI, photoplethysmogram, heart rate, blood pressures, SpO2, cardiac output, and stroke volume, were analyzed from 20 subjects before, during, and after an average controlled voluntary hemorrhage of ∼1.2 L of blood. Compensatory Reserve Index decreased by 33% in a linear fashion across progressive blood volume loss, with no clinically significant alterations in vital signs. The receiver operating characteristic area under the curve for the CRI was 0.90, with a sensitivity of 0.80 and specificity of 0.76. In comparison, blood pressures, heart rate, SpO2, cardiac output, and stroke volume had significantly lower receiver operating characteristic area under the curve values and specificities for detecting the same volume of blood loss. Consistent with our hypothesis, CRI detected blood loss and restoration with significantly greater specificity than did other traditional physiologic measures. Single measurement of CRI may enable more accurate triage, whereas CRI monitoring may allow for earlier detection of casualty deterioration.

Detection of Low-volume Blood Loss: Compensatory Reserve Versus Traditional Vital Signs

BACKGROUND Humans are able to compensate for low-volume blood loss with minimal change in traditional vital signs. We hypothesized that a novel algorithm, which analyzes photoplethysmogram (PPG) wave forms to continuously estimate compensatory reserve would provide greater sensitivity and specificity to detect low-volume blood loss compared with traditional vital signs. The compensatory reserve index (CRI) is a measure of the reserve remaining to compensate for reduced central blood volume, where a CRI of 1 represents supine normovolemia and 0 represents the circulating blood volume at which hemodynamic decompensation occurs; values between 1 and 0 indicate the proportion of reserve remaining. METHODS Subjects underwent voluntary donation of 1 U (approximately 450 mL) of blood. Demographic and continuous noninvasive vital sign wave form data were collected, including PPG, heart rate, systolic blood pressure, cardiac output, and stroke volume. PPG wave forms were later processed by the algorithm to estimate CRI values. RESULTS Data were collected from 244 healthy subjects (79 males and 165 females), with a mean (SD) age of 40.1 (14.2) years and mean (SD) body mass index of 25.6 (4.7). After blood donation, CRI significantly decreased in 92% (&agr; = 0.05; 95% confidence interval [CI], 88–95%) of the subjects. With the use of a threshold decrease in CRI of 0.05 or greater for the detection of 1 U of blood loss, the receiver operating characteristic area under the curve was 0.90, with a sensitivity of 0.84 and specificity of 0.86. In comparison, systolic blood pressure (52%; 95% CI, 45–59%), heart rate (65%; 95% CI, 58–72%), cardiac output (47%; 95% CI, 40–54%), and stroke volume (74%; 95% CI, 67–80%) changed in fewer subjects, had significantly lower receiver operating characteristic area under the curve values, and significantly lower specificities for detecting the same volume of blood loss. CONCLUSION Consistent with our hypothesis, CRI detected low-volume blood loss with significantly greater specificity than other traditional physiologic measures. These findings warrant further evaluation of the CRI algorithm in actual trauma settings. LEVEL OF EVIDENCE Diagnostic study, level II.

Early tracheostomy improves outcomes in severely injured children and adolescents

BACKGROUND Early tracheostomy has been advocated for adult trauma patients to improve outcomes and resource utilization. We hypothesized that timing of tracheostomy for severely injured children would similarly impact outcomes. METHODS Injured children undergoing tracheostomy over a 10-year period (2002-2012) were reviewed. Early tracheostomy was defined as post-injury day ≤ 7. Data were compared using Students t test, Pearson chi-squared test and Fisher exact test. Statistical significance was set at p<0.05 with 95% confidence intervals. RESULTS During the 10-year study period, 91 patients underwent tracheostomy following injury. Twenty-nine (32%) patients were < 12 years old; of these, 38% received early tracheostomy. Sixty-two (68%) patients were age 13 to 18; of these, 52% underwent early tracheostomy. Patients undergoing early tracheostomy had fewer ventilator days (p=0.003), ICU days (p=0.003), hospital days (p=0.046), and tracheal complications (p=0.03) compared to late tracheostomy. There was no difference in pneumonia (p=0.48) between early and late tracheostomy. CONCLUSION Children undergoing early tracheostomy had improved outcomes compared to those who underwent late tracheostomy. Early tracheostomy should be considered for the severely injured child. SUMMARY Early tracheostomy is advocated for adult trauma patients to improve patient comfort and resource utilization. In a review of 91 pediatric trauma patients undergoing tracheostomy, those undergoing tracheostomy on post-injury day ≤ 7 had fewer ventilator days, ICU days, hospital days, and tracheal complications compared to those undergoing tracheostomy after post-injury day 7.

Teen traffic safety campaign: competition is the key.

BACKGROUND : Motor vehicle crashes are the leading cause of death among teenagers because of higher crash rates per mile driven and low seat belt use rates. METHODS : An educational program aimed at increasing seat belt use among teens was implemented at five area high schools in the spring of 2007 and six high schools in the fall of 2007. Observational studies were conducted as students arrived at school. Resources and incentives were provided to generate peer-to-peer motivation. Schools competed against one another to see which could achieve the highest seat belt usage rate during a 7-week period. Observational studies were repeated, and success of the safety campaign was measured by an increase in seat belt usage at the participating high schools. RESULTS : At the beginning of the safety campaign, average seat belt use among teen drivers was 47% and 59% for the spring and fall, respectively. Teen passengers had an average usage of 40% and 57% for the spring and fall. In the spring campaign, seat belt use increased by 36% and 19% for teen drivers and passengers, respectively. Similar results were seen in the fall, with increases of 26% and 13% for teen drivers and passengers, respectively. Overall seat belt use in both campaigns increased by 20%, to an average use rate of 71%. CONCLUSIONS : Social pressure and poor comprehension of the risks of injury were identified as barriers to seat belt usage among teenage high school students. A friendly, competitive approach to openly discussing and educating teens about these risks led to a 20% increase in seat belt usage among teen drivers and their passengers.

Helicopter versus ground emergency medical services for the transportation of traumatically injured children

BACKGROUND Helicopter emergency medical services (HEMS) are a common mode of transportation for pediatric trauma patients. We hypothesized that HEMS improve outcomes for traumatically injured children compared to ground emergency medical services (GEMS). METHODS We queried trauma registries of two level 1 pediatric trauma centers for children 0-17 years, treated from 2003 to 2013, transported by HEMS or GEMS, with known transport starting location and outcome. A geocoding service estimated travel distance and time. Multivariate regression analyses were performed to adjust for injury severity variables and travel distance/time. RESULTS We identified 14,405 traumatically injured children; 3870 (26.9%) transported by HEMS and 10,535 (73.1%) transported by GEMS. Transport type was not significantly associated with survival, ICU length of stay, or discharge disposition. Transport by GEMS was associated with a 68.6%-53.1% decrease in hospital length of stay, depending on adjustment for distance/time. Results were similar for children with severe injuries, and with propensity score matched cohorts. Of note, 862/3850 (22.3%) of HEMS transports had an ISS<10 and hospitalization<1 day. CONCLUSIONS HEMS do not independently improve outcomes for traumatically injured children, and 22.3% of children transported by HEMS are not significantly injured. These factors should be considered when requesting HEMS for transport of traumatically injured children.

Promoting early diagnosis of hemodynamic instability during simulated hemorrhage with the use of a real-time decision-assist algorithm

BACKGROUND This study aimed to test the hypothesis that the addition of a real-time decision-assist machine learning algorithm by emergency medical system personnel could shorten the time needed to identify an unstable patient during a hemorrhage profile as compared with vital sign information alone. METHODS Fifty emergency medical team-paramedics from a large, urban fire department participated as subjects. Subjects viewed a monitor screen on two occasions as follows: (1) display of standard vital signs alone and (2) with the addition of an index (Compensatory Reserve Index) associated with estimated central blood volume status. The subjects were asked to push a computer key at any point in the sequence they believed the patient had become unstable based on information provided by the monitor screen. The average difference in time to identify hemodynamic instability between experimental and control groups was assessed by paired, two-tailed t test and reported with 95% confidence intervals (95% CI). RESULTS The mean (SD) amount of time required to identify an unstable patient was 18.3 (4.1) minutes (95% CI, 17.2–19.4 minutes) without the algorithm and 10.7 (4.2) minutes (95% CI, 9.5–11.9 minutes) with the algorithm (p < 0.001). CONCLUSION In a simulated patient encounter involving uncontrolled hemorrhage, the use of a monitor that estimates central blood volume loss was associated with early identification of impending hemodynamic instability. Physiologic monitors capable of early identification and estimation of the physiologic capacity to compensate for blood loss during hemorrhage may enable optimal guidance for hypotensive resuscitation. They may also help identify casualties benefitting from forward administration of plasma, antifibrinolytics and procoagulants in a remote damage-control resuscitation model.

The Compensatory Reserve Index Following Injury: Results of a Prospective Clinical Trial.

Introduction: Humans are able to compensate for significant blood loss with little change in traditional vital signs. We hypothesized that an algorithm, which recognizes compensatory changes in photoplethysmogram (PPG) waveforms, could detect active bleeding and ongoing volume loss in injured patients. Methods: Injured adults were prospectively enrolled at a level I trauma center. PPG data collection was conducted using a custom-made pulse oximeter. Waveform data were post-processed by an algorithm to calculate the compensatory reserve index (CRI), measured on a scale of 1 to 0, with 1 indicating fully compensated and 0 indicating no reserve, or decompensation. CRI was compared to clinical findings. Results: Fifty patients were enrolled in the study; 3 had incomplete data, 3 had indeterminate bleeding, 12 were actively bleeding, and 32 were not bleeding. The mean initial CRI of bleeding patients was significantly lower compared with the non-bleeding patients (CRI 0.17, 95% CI = 0.13–0.22 vs. CRI 0.56, 95% CI = 0.49–0.62, P < 0.001). Using a cut-off of 0.21 had a sensitivity of 0.97 and specificity of 0.83 for identifying bleeding patients. CRI had a higher sensitivity than heart rate (75%), systolic blood pressure (63%), shock index (27%), base deficit (29%), lactate (80%), hemoglobin (50%), and hematocrit (50%). During ongoing bleeding, CRI decreased following fluid resuscitation, and conversely increased for patients who were not bleeding. Conclusions: A novel computational algorithm that recognizes subtle changes in PPG waveforms can quickly and noninvasively discern which patients are actively bleeding and continuing to bleed with high sensitivity and specificity in acutely injured patients.

Emerging technologies for pediatric and adult trauma care

Purpose of review Current Emergency Medical Service protocols rely on provider-directed care for evaluation, management and triage of injured patients from the field to a trauma center. New methods to quickly diagnose, support and coordinate the movement of trauma patients from the field to the most appropriate trauma center are in development. These methods will enhance trauma care and promote trauma system development. Recent findings Recent advances in machine learning, statistical methods, device integration and wireless communication are giving rise to new methods for vital sign data analysis and a new generation of transport monitors. These monitors will collect and synchronize exponentially growing amounts of vital sign data with electronic patient care information. The application of advanced statistical methods to these complex clinical data sets has the potential to reveal many important physiological relationships and treatment effects. Summary Several emerging technologies are converging to yield a new generation of smart sensors and tightly integrated transport monitors. These technologies will assist prehospital providers in quickly identifying and triaging the most severely injured children and adults to the most appropriate trauma centers. They will enable the development of real-time clinical support systems of increasing complexity, able to provide timelier, more cost-effective, autonomous care.