Shwetha Edla

Electrocardiogram Signal Modeling With Adaptive Parameter Estimation Using Sequential Bayesian Methods

The automatic classification of electrocardiogram (ECG) signals is of great clinical significance in eliminating the strenuous process of manually annotating ECG recordings. Although statistical models describing ECG signal dynamics currently exist, they depend considerably on a priori information and user-specified model parameters. Also, ECG beat morphologies, which vary greatly across different individuals and disease states, cannot easily be described by a single representation. In this paper, we propose sequential Bayesian based methods to effectively model and adaptively select parameters of ECG signals. We first consider an adaptive framework based on a sequential Bayesian tracking method that adaptively selects the best cardiac parameters by minimizing the estimation error and does not require early-stage processing to obtain prior signal information. We then present ECG modeling techniques using the interacting multiple model (IMM) and sequential Markov chain Monte Carlo (SMCMC) methods combined with simultaneous model selection. Both these methods can adaptively choose between different representations to model various ECG beat morphologies without requiring prior ECG information. The performance of the proposed algorithms is demonstrated using real ECG data. Finally, we develop a Bayesian maximum-likelihood based classifier to classify different types of cardiac arrhythmias using which, correction classification rates of 90% and 98% are obtained, when considering features obtained from the estimated model parameters of the adaptive framework, and both the IMM and SMCMC methods, respectively.

Sequential Markov chain Monte Carlo filter with simultaneous model selection for electrocardiogram signal modeling

Constructing statistical models of electrocardiogram (ECG) signals, whose parameters can be used for automated disease classification, is of great importance in precluding manual annotation and providing prompt diagnosis of cardiac diseases. ECG signals consist of several segments with different morphologies (namely the P wave, QRS complex and the T wave) in a single heart beat, which can vary across individuals and diseases. Also, existing statistical ECG models exhibit a reliance upon obtaining a priori information from the ECG data by using preprocessing algorithms to initialize the filter parameters, or to define the user-specified model parameters. In this paper, we propose an ECG modeling technique using the sequential Markov chain Monte Carlo (SMCMC) filter that can perform simultaneous model selection, by adaptively choosing from different representations depending upon the nature of the data. Our results demonstrate the ability of the algorithm to track various types of ECG morphologies, including intermittently occurring ECG beats. In addition, we use the estimated model parameters as the feature set to classify between ECG signals with normal sinus rhythm and four different types of arrhythmia.

Electrocardiogram signal modeling using interacting multiple models

The automatic classification of different heart diseases for monitoring cardiac health through the use of dynamic modeling of electrocardiogram (ECG) signals would yield innovative findings of immense clinical importance. This has been a difficult problem, however, as ECG signals consist of fiducial points with different morphologies within a single heart beat; the points vary between persons and disease states and cannot be described by a single representation. Current statistical ECG models depend on user-specified parameters and a priori information that requires pre-processing. In this paper, we propose a novel method for dynamically modeling, estimating and classifying ECG signals by representing different heart diseases using the interacting multiple model (IMM) algorithm, which can adaptively choose between different representations depending on the ECG data morphology. Using real ECG signals, we demonstrate that the IMM-based model can accurately represent different morphologies with minimal prior information. Using the estimated model parameters as a low-dimensional feature set, we also showed high classification performance between different cardiac arrhythmias.

Is heart rate variability better than routine vital signs for prehospital identification of major hemorrhage

OBJECTIVE During initial assessment of trauma patients, metrics of heart rate variability (HRV) have been associated with high-risk clinical conditions. Yet, despite numerous studies, the potential of HRV to improve clinical outcomes remains unclear. Our objective was to evaluate whether HRV metrics provide additional diagnostic information, beyond routine vital signs, for making a specific clinical assessment: identification of hemorrhaging patients who receive packed red blood cell (PRBC) transfusion. METHODS Adult prehospital trauma patients were analyzed retrospectively, excluding those who lacked a complete set of reliable vital signs and a clean electrocardiogram for computation of HRV metrics. We also excluded patients who did not survive to admission. The primary outcome was hemorrhagic injury plus different PRBC transfusion volumes. We performed multivariate regression analysis using HRV metrics and routine vital signs to test the hypothesis that HRV metrics could improve the diagnosis of hemorrhagic injury plus PRBC transfusion vs routine vital signs alone. RESULTS As univariate predictors, HRV metrics in a data set of 402 subjects had comparable areas under receiver operating characteristic curves compared with routine vital signs. In multivariate regression models containing routine vital signs, HRV parameters were significant (P<.05) but yielded areas under receiver operating characteristic curves with minimal, nonsignificant improvements (+0.00 to +0.05). CONCLUSIONS A novel diagnostic test should improve diagnostic thinking and allow for better decision making in a significant fraction of cases. Our findings do not support that HRV metrics add value over routine vital signs in terms of prehospital identification of hemorrhaging patients who receive PRBC transfusion.

Adaptive parameter estimation of cardiovascular signals using sequential Bayesian techniques

Parameter estimation of biological signals such as the electrocardiogram (ECG) is of key clinical significance and can be used to monitor cardiac health and diagnose heart diseases. However, statistical ECG models with unknown parameters depend upon a priori parameters such as mean cardiac frequency and user-specified parameters such as the number of harmonics in the ECG model. These parameters can vary from patient to patient and with different disease stages. In this paper, we propose a sequential Bayesian tracking method to adaptively select the best cardiac parameters in order to minimize the parameter estimation error. Our results using real ECG data demonstrate the importance of the adaptive algorithm for selecting cardiac parameters at each time instant and show how these parameters can be used to classify different types of ECG signals.

Muscle Oxygen Saturation Improves Diagnostic Association Between Initial Vital Signs and Major Hemorrhage: A Prospective Observational Study.

Abstract Objectives During initial assessment of trauma patients, vital signs do not identify all patients with life‐threatening hemorrhage. We hypothesized that a novel vital sign, muscle oxygen saturation (SmO2), could provide independent diagnostic information beyond routine vital signs for identification of hemorrhaging patients who require packed red blood cell (RBC) transfusion. Methods This was an observational study of adult trauma patients treated at a Level I trauma center. Study staff placed the CareGuide 1100 tissue oximeter (Reflectance Medical Inc., Westborough, MA), and we analyzed average values of SmO2, systolic blood pressure (sBP), pulse pressure (PP), and heart rate (HR) during 10 minutes of early emergency department evaluation. We excluded subjects without a full set of vital signs during the observation interval. The study outcome was hemorrhagic injury and RBC transfusion ≥ 3 units in 24 hours (24‐hr RBC ≥ 3). To test the hypothesis that SmO2 added independent information beyond routine vital signs, we developed one logistic regression model with HR, sBP, and PP and one with SmO2 in addition to HR, sBP, and PP and compared their areas under receiver operating characteristic curves (ROC AUCs) using DeLongs test. Results We enrolled 487 subjects; 23 received 24‐hr RBC ≥ 3. Compared to the model without SmO2, the regression model with SmO2 had a significantly increased ROC AUC for the prediction of ≥ 3 units of 24‐hr RBC volume, 0.85 (95% confidence interval [CI], 0.75–0.91) versus 0.77 (95% CI, 0.66–0.86; p < 0.05 per DeLongs test). Results were similar for ROC AUCs predicting patients (n = 11) receiving 24‐hr RBC ≥ 9. Conclusions SmO2 significantly improved the diagnostic association between initial vital signs and hemorrhagic injury with blood transfusion. This parameter may enhance the early identification of patients who require blood products for life‐threatening hemorrhage.

A Platform for Real-time Acquisition and Analysis of Physiological Data in Hospital Emergency Departments

An opportunity exists for automated clinical decision support, in which raw source data from a conventional physiological monitoring system are continuously streamed to an independent analysis platform. Such a system would enable a wider range of functionality than offered by the source monitoring system. Although vendor solutions for this purpose are emerging, we developed our own system in order to control the expense and to permit forensic analysis of the internal core functionality of the system. In this report, we describe a platform that can provide decision support for trauma patients in an Emergency Department (ED). System evaluation spanned 39 days, and included a total of 2200 patient session hrs of real-time monitoring. We highlight the technical issues that we confronted, including protection of the core monitoring network, the real-time communication of electronic medical data, and the reliability of the real-time analysis. Detailing these nuanced technical issues may be valuable to other software developers or for those interested in investing in a vendor solution for similar functionality.

Tachycardic and non-tachycardic responses in trauma patients with haemorrhagic injuries

BACKGROUND Analyses of large databases have demonstrated that the association between heart rate (HR) and blood loss is weaker than what is taught by Advanced Trauma Life Support training. However, those studies had limited ability to generate a more descriptive paradigm, because they only examined a single HR value per patient. METHODS In a comparative, retrospective analysis, we studied the temporal characteristics of HR through time in adult trauma patients with haemorrhage, based on documented injuries and transfusion of ≥3 units of red blood cells (RBCs). We analysed archived vital-sign data of up to 60 min during either pre-hospital or emergency department care. RESULTS We identified 133 trauma patients who met the inclusion criteria for major haemorrhage and 1640 control patients without haemorrhage. There were 55 haemorrhage patients with a normal median HR and 78 with tachycardia. Median ΔHR was -0.8 and +0.7 bpm per 10 min, respectively. Median time to documented hypotension was 8 and 5 min, respectively. RBCs were not significantly different; median volumes were 6 (IQR: 4-13) and 10 units (IQR: 5-16), respectively. Time-to-hypotension and mortality were not significantly different. Tachycardic patients were significantly younger (P < 0.05). Only 10 patients with normal HR developed transient/temporary tachycardia, and only 11 tachycardic patients developed a transient/temporary normal HR. CONCLUSIONS The current analysis suggests that some trauma patients with haemorrhage are continuously tachycardic while others have a normal HR. For both cohorts, hypotension typically develops within 30 min, without any consistent temporal increases or trends in HR.

The matching of sinus arrhythmia to respiration: are trauma patients without serious injury comparable to healthy laboratory subjects?

We sought to better understand the physiology underlying the metrics of heart rate variability (HRV) in trauma patients without serious injury, compared to healthy laboratory controls. In trauma patients without serious injury (110 subjects, 470 2-min data segments), we studied the correlation between sinus arrhythmia (SA) rate, heart rate (HR), and respiratory rate (RR). Most segments with 2.4 <; HR/RR <; 4.8 exhibited SA-RR matching, whereas rate matching was absent in 81% of the segments with HR/RR <; 2.4 and in 86% of the segments with HR/RR > 4.8. The findings were comparable, in some cases remarkably so, to previous reports from healthy laboratory subjects. The presence (or absence) of SA-RR matching, when SA is largely controlled by respiration, can be anticipated in this trauma population. This work provides a valuable step towards the definition of patterns of HRV found in trauma patients with and without life-threatening injury.

A Comparison of Alerting Strategies for Hemorrhage Identification during Prehospital Emergency Transport

Early and accurate identification of physiological abnormalities is one feature of intelligent decision support. The ideal analytic strategy for identifying pathological states would be highly sensitive and highly specific, with minimal latency. In the field of manufacturing, there are well-established analytic strategies for statistical process control, whereby aberrancies in a manufacturing process are detected by monitoring and analyzing the process output. These include simple thresholding, the sequential probability ratio test (SPRT), risk-adjusted SPRT, and the cumulative sum method. In this report, we applied these strategies to continuously monitored prehospital vital-sign data from trauma patients during their helicopter transport to level I trauma centers, seeking to determine whether one strategy would be superior. We found that different configurations of each alerting strategy yielded widely different performances in terms of sensitivity, specificity, and average time to alert. Yet, comparing the different investigational analytic strategies, we observed substantial overlap among their different configurations, without any one analytic strategy yielding distinctly superior performance. In conclusion, performance did not depend as much on the specific analytic strategy as much as the configuration of each strategy. This implies that any analytic strategy must be carefully configured to yield the optimal performance (i.e., the optimal balance between sensitivity, specificity, and latency) for a specific use case. Conversely, this also implies that an alerting strategy optimized for one use case (e.g., long prehospital transport times) may not necessarily yield performance data that are optimized for another clinical application (e.g., short prehospital transport times, intensive care units, etc.).