An annotated ventricular tachycardia (VT) alarm database: Toward a uniform standard for optimizing automated VT identification in hospitalized patients.

BACKGROUND: False ventricular tachycardia (VT) alarms are common during in-hospital electrocardiographic (ECG) monitoring. Prior research shows that the majority of false VT can be attributed to algorithm deficiencies. PURPOSE: The purpose of this study was: (1) to describe the creation of a VT database annotated by ECG experts and (2) to determine true vs. false VT using a new VT algorithm created by our group. METHODS: The VT algorithm was processed in 5320 consecutive ICU patients with 572,574 h of ECG and physiologic monitoring. A search algorithm identified potential VT, defined as: heart rate >100 beats/min, QRSs > 120 ms, and change in QRS morphology in >6 consecutive beats compared to the preceding native rhythm. Seven ECG channels, SpO2 , and arterial blood pressure waveforms were processed and loaded into a web-based annotation software program. Five PhD-prepared nurse scientists performed the annotations. RESULTS: Of the 5320 ICU patients, 858 (16.13%) had 22,325 VTs. After three levels of iterative annotations, a total of 11,970 (53.62%) were adjudicated as true, 6485 (29.05%) as false, and 3870 (17.33%) were unresolved. The unresolved VTs were concentrated in 17 patients (1.98%). Of the 3870 unresolved VTs, 85.7% (n = 3281) were confounded by ventricular paced rhythm, 10.8% (n = 414) by underlying BBB, and 3.5% (n = 133) had a combination of both. CONCLUSIONS: The database described here represents the single largest human-annotated database to date. The database includes consecutive ICU patients, with true, false, and challenging VTs (unresolved) and could serve as a gold standard database to develop and test new VT algorithms.

Ischemic Pre-Conditioning is Present in Patients with Non-ST Elevation Myocardial Infarction and ECG Derived Moderate Obstructive Sleep Apnea.

Background: Obstructive Sleep Apnea (OSA) is associated with an increased risk of cardiovascular events, including Acute Coronary Syndrome (ACS). There is conflicting evidence that suggests OSA has a cardioprotective effect (i.e., lower troponin), via ischemic pre-conditioning, in patients with ACS. Purpose: This study had two aims: (1) compare peak troponin between non-ST elevation (NSTE) ACS patients with and without moderate OSA identified using a Holter derived respiratory disturbance index (HDRDI); and (2) determine the frequency of transient myocardial ischemia (TMI) between NSTE-ACS patients with and without moderate HDRDI. Method: This was a secondary analysis. OSA events were identified from 12-lead ECG Holter recordings using QRSs, R-R intervals, and the myogram. Moderate OSA was defined as an HDRDI ≥15 events per/hour. TMI was defined as ≥1 millimeter of ST-segment ↑ or ↓, in ≥ 1 ECG lead, ≥ 1 minute. Results: In 110 NSTE-ACS patients, 39% (n=43) had moderate HDRDI. Peak troponin was higher in patients with moderate HDRDI (6.8 ng/ml yes vs. 10.2 ng/ml no; p=0.037). There was a trend for fewer TMI events, but there were no differences (16% yes vs. 30% no; p=0.081). Conclusions: NSTE-ACS patients with moderate HDRDI have less cardiac injury than those without moderate HDRDI measured using a novel ECG derived method. These findings corroborate prior studies suggesting a possible cardioprotective effect of OSA in ACS patients via ischemic pre-condition. There was a trend for fewer TMI events in moderate HDRDI patients, but there was no statistical difference. Future research should explore the underlying physiologic mechanisms of this finding.

Agreement between respiratory rate measurement using a combined electrocardiographic derived method versus impedance from pneumography.

BACKGROUND: Impedance pneumography (IP) is the current device-driven method used to measure respiratory rate (RR) in hospitalized patients. However, RR alarms are common and contribute to alarm fatigue. While RR derived from electrocardiographic (ECG) waveforms hold promise, they have not been compared to the IP method. PURPOSE: Study examined the agreement between the IP and combined-ECG derived (EDR) for normal RR (≥12 or ≤20 breaths/minute [bpm]); low RR (≤5 bpm); and high RR (≥30 bpm). METHODOLOGY: One-hundred intensive care unit patients were included by RR group: (1) normal RR (n = 50; 25 low RR and 25 high RR); (2) low RR (n = 50); and (3) high RR (n = 50). Bland-Altman analysis was used to evaluate agreement. RESULTS: For normal RR, a significant bias difference of -1.00 + 2.11 (95% CI -1.60 to -0.40) and 95% limit of agreement (LOA) of -5.13 to 3.13 was found. For low RR, a significant bias difference of -16.54 + 6.02 (95% CI: -18.25 to -14.83) and a 95% LOA of -28.33 to - 4.75 was found. For high RR, a significant bias difference of 17.94 + 12.01 (95% CI: 14.53 to 21.35) and 95% LOA of -5.60 to 41.48 was found. CONCLUSION: Combined-EDR method had good agreement with the IP method for normal RR. However, for the low RR, combined-EDR was consistently higher than the IP method and almost always lower for the high RR, which could reduce the number of RR alarms. However, replication in a larger sample including confirmation with visual assessment is warranted.

Detection of sleep-disordered breathing with ambulatory Holter monitoring.

PURPOSE: Obstructive sleep apnea (OSA) syndrome is a common condition that can impact clinical outcomes among patients with cardiovascular disease. Screening all subjects with heart disease via polysomnography (PSG) is costly and resource-limited. We sought to compare a Holter monitor-based algorithm to detect OSA to in-laboratory polysomnography (PSG). METHODS: Prospective cohort study of patients undergoing in-laboratory attended PSG for the evaluation of OSA. A standard 12-lead Holter monitor was attached to patients at the initiation of PSG. Holter-derived respiratory disturbance index (HDRDI) was extracted from the respiratory myogram, based on detecting skeletal muscle "noise" detected on the baseline. Apneic and hypopneic episodes were identified by comparing sudden changes in the myogram to abrupt increases in heart rate. The HDRDI was compared with the PSG-derived apnea-hypopnea index (PDAHI). RESULTS: Thirty patients underwent simultaneous Holter monitoring and overnight diagnostic PSG. An ROC curve for peak HDRDI was 0.79 (95% CI 0.61, 0.97) for OSA, with sensitivity of 94.4% and specificity of 54.5%. A cutoff value of HDRDI < 10 appeared to identify those individuals without clinically significant sleep-disordered breathing. CONCLUSION: Holter-derived respiration detected OSA comparable to PSG. Further study is warranted to determine its utility for screening and diagnosing OSA in appropriately selected patients.

Evaluation of ECG algorithms designed to improve detect of transient myocardial ischemia to minimize false alarms in patients with suspected acute coronary syndrome.

BACKGROUND: Patients hospitalized for suspected acute coronary syndrome (ACS) are at risk for transient myocardial ischemia. During the "rule-out" phase, continuous ECG ST-segment monitoring can identify transient myocardial ischemia, even when asymptomatic. However, current ST-segment monitoring software is vastly underutilized due to false positive alarms, with resultant alarm fatigue. Current ST algorithms may contribute to alarm fatigue because; (1) they are not designed with a delay (minutes), rather alarm to brief spikes (i.e., turning, heart rate changes), and (2) alarm to changes in a single ECG lead, rather than contiguous leads. PURPOSE: This study was designed to determine sensitivity, and specificity, of ST algorithms when accounting for; ST magnitude (100µV vs 200µV), duration, and changes in contiguous ECG leads (i.e., aVL, I, - aVR, II, aVF, III; V1, V2, V3, V4, V5, V6, V6, I). METHODS: This was a secondary analysis from the COMPARE Study, which assessed occurrence rates for transient myocardial ischemia in hospitalized patients with suspected ACS using 12-lead Holter. Transient myocardial ischemia was identified from Holter using >100µV ST-segment ↑ or ↓, in >1 ECG lead, >1min. Algorithms tested against Holter transient myocardial ischemia were done using the University of California San Francisco (UCSF) ECG algorithm and included: (1)100µV vs 200µV any lead during a 5-min ST average; (2)100µV vs 200µV any lead >5min, (3) 100µV vs 200µV any lead during a 5-min ST average in contiguous leads, and (4) 100µV vs 200µV>5min in contiguous leads (Table below). RESULTS: In 361 patients; mean age 63+12years, 63% male, 56% prior CAD, 43 (11%) had transient myocardial ischemia. Of the 43 patients with transient myocardial ischemia, 17 (40%) had ST-segment elevation events, and 26 (60%) ST-segment depression events. A higher proportion of patients with ST segment depression has missed ischemic events. Table shows sensitivity and specificity for the four algorithms tested. CONCLUSIONS: Sensitivity was highly variable, due to the ST threshold selected, with the 100µV measurement point being superior to the 200µV amplitude threshold. Of all the algorithms tested, there was moderate sensitivity and specificity (70% and 68%) using the 100µV ST-segment threshold, integrated ST-segment changes in contiguous leads during a 5-min average.

ECG-derived Cheyne-Stokes respiration and periodic breathing in healthy and hospitalized populations.

BACKGROUND: Cheyne-Stokes respiration (CSR) has been investigated primarily in outpatients with heart failure. In this study we compare CSR and periodic breathing (PB) between healthy and cardiac groups. METHODS: We compared CSR and PB, measured during 24 hr of continuous 12-lead electrocardiographic (ECG) Holter recording, in a group of 90 hospitalized patients presenting to the emergency department with symptoms suggestive of acute coronary syndrome (ACS) to a group of 100 healthy ambulatory participants. We also examined CSR and PB in the 90 patients presenting with ACS symptoms, divided into a group of 39 (43%) with confirmed ACS, and 51 (57%) with a cardiac diagnosis but non-ACS. SuperECG software was used to derive respiration and then calculate CSR and PB episodes from the ECG Holter data. Regression analyses were used to analyze the data. We hypothesized SuperECG software would differentiate between the groups by detecting less CSR and PB in the healthy group than the group of patients presenting to the emergency department with ACS symptoms. RESULTS: Hospitalized patients with suspected ACS had 7.3 times more CSR episodes and 1.6 times more PB episodes than healthy ambulatory participants. Patients with confirmed ACS had 6.0 times more CSR episodes and 1.3 times more PB episodes than cardiac non-ACS patients. CONCLUSION: Continuous 12-lead ECG derived CSR and PB appear to differentiate between healthy participants and hospitalized patients.

The J to T-peak interval as a biomarker in drug safety studies: A method of accuracy assessment applied to two algorithms.

OBJECTIVES: To evaluate performance of J-to-T-peak (JTP) measurements of 12-lead ECGs, in a five-arm study using drugs with various levels of electrolyte channel block. METHODS: The novel evaluation method distinguishes between different aspects of measurement. "Random noise" is the variability among repeated measurements made without changing the conditions. "Context noise" is the variability of changes in context of the measurement, e.g. T-wave morphology, autonomic nervous system state. RESULTS: The average random noise of our RR-corrected JTPc measurements in standard deviations was 3.0 ms and not dependent on the drug. The average context noise was 4.0 ms for ranolazine, verapamil, and placebo, and 8.8 ms for dofetilide and quinidine. Measurement consistency is corroborated by linear fit confidence intervals of baseline- and placebo-corrected JTPc versus drug concentration. CONCLUSIONS: Systematic differences were found in JTPc drug response between the Mortara method and published data. Residual signal component in the context noise may influence future study design.

Evaluation of beat-to-beat ventricular repolarization lability from standard 12-lead ECG during acute myocardial ischemia.

BACKGROUND: Acute myocardial ischemia is a common cause of ventricular arrhythmias, yet recent ECG methods predicting susceptibility to ventricular tachyarrhythmia have not been fully evaluated during spontaneous ischemia. We sought to evaluate the clinical utility of alternans and non-alternans components of repolarization variability from the standard 10-second 12-lead ECG signals to risk stratify patients with acute chest pain. METHODS: We enrolled consecutive, non-traumatic, chest pain patients transported through Emergency Medical Services (EMS) to three tertiary care hospitals with cardiac catheterization lab capabilities in Pittsburgh, PA. ECG signals were manually annotated by an electrophysiologist, then automatically processed using a custom-written software. Both T wave alternans (TWA) and non-alternans repolarization variability (NARV) were calculated using the absolute RMS differences over the repolarization window between odd/even averaged beats and between consecutive averaged pairs, respectively. The primary study outcome was the presence of acute myocardial infarction (AMI) documented by cardiac angiography. RESULTS: After excluding patients with secondary repolarization changes (n=123) and those with excessive noise (n=90), our final sample included 537 patients (age 57±16years, 56% males). Patients with AMI (n=47, 9%) had higher TWA and NARV values (p<0.01). Mean RR correlated with TWA, and noise measures correlated with TWA and NARV, after adjusting for potential confounders. There was a high collinearity between TWA and NARV, and each was separately predictive of AMI after controlling for number of analyzed beats, noise measures, and other clinical variables. CONCLUSIONS: Despite limitations imposed by signal quality, TWA and NARV are higher in patients with AMI, even after correction for potential confounders. The clinical value of TWA and NARV derived from standard ECG using our time-domain RMS method is questionable due to the small number of beats and significant noise.

Comparison of automated measurements of electrocardiographic intervals and durations by computer-based algorithms of digital electrocardiographs.

BACKGROUND AND PURPOSE: Automated measurements of electrocardiographic (ECG) intervals are widely used by clinicians for individual patient diagnosis and by investigators in population studies. We examined whether clinically significant systematic differences exist in ECG intervals measured by current generation digital electrocardiographs from different manufacturers and whether differences, if present, are dependent on the degree of abnormality of the selected ECGs. METHODS: Measurements of RR interval, PR interval, QRS duration, and QT interval were made blindly by 4 major manufacturers of digital electrocardiographs used in the United States from 600 XML files of ECG tracings stored in the US FDA ECG warehouse and released for the purpose of this study by the Cardiac Safety Research Consortium. Included were 3 groups based on expected QT interval and degree of repolarization abnormality, comprising 200 ECGs each from (1) placebo or baseline study period in normal subjects during thorough QT studies, (2) peak moxifloxacin effect in otherwise normal subjects during thorough QT studies, and (3) patients with genotyped variants of congenital long QT syndrome (LQTS). RESULTS: Differences of means between manufacturers were generally small in the normal and moxifloxacin subjects, but in the LQTS patients, differences of means ranged from 2.0 to 14.0 ms for QRS duration and from 0.8 to 18.1 ms for the QT interval. Mean absolute differences between algorithms were similar for QRS duration and QT intervals in the normal and in the moxifloxacin subjects (mean ≤6 ms) but were significantly larger in patients with LQTS. CONCLUSIONS: Small but statistically significant group differences in mean interval and duration measurements and means of individual absolute differences exist among automated algorithms of widely used, current generation digital electrocardiographs. Measurement differences, including QRS duration and the QT interval, are greatest for the most abnormal ECGs.

The development and validation of an early warning system to prevent the acquisition of 12-lead resting ECGs with interchanged electrode positions.

BACKGROUND: A fraction of routine resting ECG's are taken with electrode positions interchanged, leading to possible clinical misinterpretation. OBJECTIVE: Develop and test a method to detect and prevent electrode reversals at the electrocardiograph before the ECG is acquired. METHOD: The algorithm is based on QRS axis and P amplitudes for limb electrode reversals, and P-Q-RS amplitude distances to detect chest electrode reversals. The evaluation method involved a large (>18,000) hospital database for which serial ECG's were available and was based on simulated juxtapositions. RESULTS: The 7 most common lead reversals could be detected with a specificity of 99.8% per type and an average sensitivity of 90%, excluding LA-LL reversal (22% sensitivity). DISCUSSION: Results are similar to retrospective studies that used smaller, more homogeneous datasets. CONCLUSION: The early warning system reduces the ECG's recorded with reversal by 80%, at the price of a modest false alert rate of 1.4%.

Assessing the interaction of respiration and heart rate in heart failure and controls using ambulatory Holter recordings.

Breathing is a critical component of cardiopulmonary function, but few tools exist to evaluate respiration in ambulatory patients. Holter monitoring allows accurate diagnosis of a host of cardiac issues, and several investigators have demonstrated the ability to detect respiratory effort on the electrocardiogram. In this study we introduce a myogram signal derived from 12-lead, high frequency Holter as a means of detecting respiratory effort. Using the combined myogram and ECG signal, four novel variables were created: total number of Cheyne-Stokes episodes; the BWRatio, the ratio of power (above baseline) measured one second after peak-to-peak respiratory power, an assessment of the "shape" of the respiratory effort; DRR, the change in RR interval centering around peak inspiration; and minutes of synchronized breathing, a fixed ratio of heart beats to respiratory cycles. These variables were assessed in 24-hour recordings from three cohorts: healthy volunteers (n=33), heart failure subjects from the GISSI HF trial (n=383), and subjects receiving implantable defibrillators with severely depressed left ventricular function enrolled in the M2Risk trial (n=470). We observed a statistically significant 6-fold increase in the number of Cheyne-Stokes episodes (p=0.01 by ANOVA), decreases in BWRatio (p<0.001), as well as decrease in DRR in heart failure subjects; only minutes of synchronized breathing was not significantly decreased in heart failure. This study provides "proof of concept" that novel variables incorporating Holter-derived respiration can distinguish healthy subjects from heart failure. The utility of these variables for predicting heart failure, arrhythmia, and death risk in prospective studies needs to be assessed.

Continuous Atrial Fibrillation Monitoring From Photoplethysmography: Comparison Between Supervised Deep Learning and Heuristic Signal Processing.

BACKGROUND: Continuous monitoring for atrial fibrillation (AF) using photoplethysmography (PPG) from smartwatches or other wearables is challenging due to periods of poor signal quality during motion or suboptimal wearing. As a result, many consumer wearables sample infrequently and only analyze when the user is at rest, which limits the ability to perform continuous monitoring or to quantify AF. OBJECTIVES: This study aimed to compare 2 methods of continuous monitoring for AF in free-living patients: a well-validated signal processing (SP) heuristic and a convolutional deep neural network (DNN) trained on raw signal. METHODS: We collected 4 weeks of continuous PPG and electrocardiography signals in 204 free-living patients. Both SP and DNN models were developed and validated both on holdout patients and an external validation set. RESULTS: The results show that the SP model demonstrated receiver-operating characteristic area under the curve (AUC) of 0.972 (sensitivity 99.6%, specificity: 94.4%), which was similar to the DNN receiver-operating characteristic AUC of 0.973 (sensitivity 92.2, specificity: 95.5%); however, the DNN classified significantly more data (95% vs 62%), revealing its superior tolerance of tracings prone to motion artifact. Explainability analysis revealed that the DNN automatically suppresses motion artifacts, evaluates irregularity, and learns natural AF interbeat variability. The DNN performed better and analyzed more signal in the external validation cohort using a different population and PPG sensor (AUC, 0.994; 97% analyzed vs AUC, 0.989; 88% analyzed). CONCLUSIONS: DNNs perform at least as well as SP models, classify more data, and thus may be better for continuous PPG monitoring.

Ventricular tachycardia and in-hospital mortality in the intensive care unit.

Background: Continuous electrocardiographic (ECG) monitoring is used to identify ventricular tachycardia (VT), but false alarms occur frequently. Objective: The purpose of this study was to assess the rate of 30-day in-hospital mortality associated with VT alerts generated from bedside ECG monitors to those from a new algorithm among intensive care unit (ICU) patients. Methods: We conducted a retrospective cohort study in consecutive adult ICU patients at an urban academic medical center and compared current bedside monitor VT alerts, VT alerts from a new-unannotated algorithm, and true-annotated VT. We used survival analysis to explore the association between VT alerts and mortality. Results: We included 5679 ICU admissions (mean age 58 ± 17 years; 48% women), 503 (8.9%) experienced 30-day in-hospital mortality. A total of 30.1% had at least 1 current bedside monitor VT alert, 14.3% had a new-unannotated algorithm VT alert, and 11.6% had true-annotated VT. Bedside monitor VT alert was not associated with increased rate of 30-day mortality (adjusted hazard ratio [aHR] 1.06; 95% confidence interval [CI] 0.88-1.27), but there was an association for VT alerts from our new-unannotated algorithm (aHR 1.38; 95% CI 1.12-1.69) and true-annotated VT(aHR 1.39; 95% CI 1.12-1.73). Conclusion: Unannotated and annotated-true VT were associated with increased rate of 30-day in-hospital mortality, whereas current bedside monitor VT was not. Our new algorithm may accurately identify high-risk VT; however, prospective validation is needed.

Time dependent history improves QT interval estimation in atrial fibrillation.

PURPOSE: It is not recommended to perform QTc estimation in patients with atrial fibrillation (AF). We evaluated multiple QT interval correction formulas, including a novel time-dependent history approach, in an effort to identify the best method for correcting the QT interval in patients with AF. The ideal correction results in independence between the QTc estimate and HR. METHODS: Per-beat characteristics were derived using SuperECG (Mortara Instrument). Offline beat-to-beat QTc interval estimates were constructed using standard formulae and averaged (2-10) groups constructed. RESULTS: Seventy-one patients were included, age 67 ± 10 years, 69% men. Mean-mean QTc intervals varied by correction (range 394-459 ms). Averaging resulted in the same mean-mean QTc estimate, but significantly reduced variability by up to 55%. Time-dependent RR interval history reduced variability the most (Δ 80%), increased QT/RR dynamics (m=.03 vs .17), and was independent with HR (m = 0.0008). CONCLUSIONS: Our data suggests that QTc interval estimation in patients with AF can be performed reliably using time-dependent history (RRc) outperforming other correction methods.

Validation of an algorithm for continuous monitoring of atrial fibrillation using a consumer smartwatch.

BACKGROUND: Consumer devices with broad reach may be useful in screening for atrial fibrillation (AF) in appropriate populations. However, currently no consumer devices are capable of continuous monitoring for AF. OBJECTIVE: The purpose of this study was to estimate the sensitivity and specificity of a smartwatch algorithm for continuous detection of AF from sinus rhythm in a free-living setting. METHODS: We studied a commercially available smartwatch with photoplethysmography (W-PPG) and electrocardiogram (W-ECG) capabilities. We validated a novel W-PPG algorithm combined with a W-ECG algorithm in a free-living setting, and compared the results to those of a 28-day continuous ECG patch (P-ECG). RESULTS: A total of 204 participants completed the free-living study, recording 81,944 hours with both P-ECG and smartwatch measurements. We found sensitivity of 87.8% (95% confidence interval [CI] 83.6%-91.0%) and specificity of 97.4% (95% CI 97.1%-97.7%) for the W-PPG algorithm (every 5-minute classification); sensitivity of 98.9% (95% CI 98.1%-99.4%) and specificity of 99.3% (95% CI 99.1%-99.5%) for the W-ECG algorithm; and sensitivity of 96.9% (95% CI 93.7%-98.5%) and specificity of 99.3% (95% CI 98.4%-99.7%) for W-PPG triggered W-ECG with a single W-ECG required for confirmation of AF. We found a very strong correlation of W-PPG in quantifying AF burden compared to P-ECG (r = 0.98). CONCLUSION: Our findings demonstrate that a novel algorithm using a commercially available smartwatch can continuously detect AF with excellent performance and that confirmation with W-ECG further enhances specificity. In addition, our W-PPG algorithm can estimate AF burden. Further research is needed to determine whether this algorithm is useful in screening for AF in select at-risk patients.

Automated QT measurement and application to detection of moxifloxacin-induced changes.

BACKGROUND: Concern for drug-induced QT prolongation has caused significant investment in QT measurement to safety-test new compounds. Manual methods are expensive and time-consuming. Reliable automatic methods would be highly desirable. METHODS: Twelve-lead Holter recordings were annotated beat-to-beat by an automatic algorithm for global QRS onset and T offset. T offset was established from the time of peak T downslope plus a rate-dependent offset, analogous to the "tangent method," wherein T offset is determined by extrapolating the T downslope to an intersection with the baseline. RESULTS AND CONCLUSIONS: Variances of the beat-to-beat QT measurements were in the range 2.5-3.4 ms over three distinct databases, including a large heart failure database. Application to a moxifloxacin/placebo control database of 29 subjects showed excellent results.

ECG derived Cheyne-Stokes respiration and periodic breathing are associated with cardiorespiratory arrest in intensive care unit patients.

BACKGROUND: Cheyne-Stokes respiration and periodic breathing (CSRPB) have not been studied sufficiently in the intensive care unit setting (ICU). OBJECTIVES: To determine whether CSRPB is associated with adverse outcomes in ICU patients. METHODS: The ICU group was divided into quartiles by CSRPB (86 patients in quartile 1 had the least CSRPB and 85 patients in quartile 4 had the most CSRPB). Adverse outcomes (emergent intubation, cardiorespiratory arrest, inpatient mortality and the composite of all) were compared between patients with most CSRPB (quartile 4) and those with least CSRPB (quartile 1). RESULTS: ICU patients in quartile 4 had a higher proportion of cardiorespiratory arrests (5% versus 0%, (p=.042), and more adverse events over all (19% versus 8%, p=.041) as compared to patients in quartile 1. CONCLUSIONS: CSRPB can be measured in the ICU and it's severity is associated with adverse outcomes in critically ill patients.

Usefulness of Trends in Continuous Electrocardiographic Telemetry Monitoring to Predict In-Hospital Cardiac Arrest.

Survival from in-hospital cardiac arrest (IHCA) due to pulseless electrical activity/asystole remains poor. We aimed to evaluate whether electrocardiographic changes provide predictive information for risk of IHCA from pulseless electrical activity/asystole. We conducted a retrospective case-control study, utilizing continuous electrocardiographic data from case and control patients. We selected 3 consecutive 3-hour blocks (block 3, 2, and 1 in that order); block 1 immediately preceded cardiac arrest in cases, whereas block 1 was chosen at random in controls. In each block, we measured dominant positive and negative trends in electrocardiographic parameters, evaluated for arrhythmias, and compared these between consecutive blocks. We created random forest and logistic regression models, and tested them on differentiating case versus control patients (case block 1 vs control block 1), and temporal relation to cardiac arrest (case block 2 vs case block 1). Ninety-one cases (age 63.0 ± 17.6, 58% male) and 1,783 control patients (age 63.5 ± 14.8, 67% male) were evaluated. We found significant differences in electrocardiographic trends between case and control block 1, particularly in QRS duration, QTc, RR, and ST. New episodes of atrial fibrillation and bradyarrhythmias were more common before IHCA. The optimal model was the random forest, achieving an area under the curve of 0.829, 63.2% sensitivity, 94.6% specificity at differentiating case versus control block 1 on a validation set, and area under the curve 0.954, 91.2% sensitivity, 83.5% specificity at differentiating case block 1 versus case block 2. In conclusion, trends in electrocardiographic parameters during the 3-hour window immediately preceding IHCA differ significantly from other time periods, and provide robust predictive information.

A Deep Learning Approach to Examine Ischemic ST Changes in Ambulatory ECG Recordings.

Patients with suspected acute coronary syndrome (ACS) are at risk of transient myocardial ischemia (TMI), which could lead to serious morbidity or even mortality. Early detection of myocardial ischemia can reduce damage to heart tissues and improve patient condition. Significant ST change in the electrocardiogram (ECG) is an important marker for detecting myocardial ischemia during the rule-out phase of potential ACS. However, current ECG monitoring software is vastly underused due to excessive false alarms. The present study aims to tackle this problem by combining a novel image-based approach with deep learning techniques to improve the detection accuracy of significant ST depression change. The obtained convolutional neural network (CNN) model yields an average area under the curve (AUC) at 89.6% from an independent testing set. At selected optimal cutoff thresholds, the proposed model yields a mean sensitivity at 84.4% while maintaining specificity at 84.9%.