Machine learning for ECG diagnosis and risk stratification of occlusion myocardial infarction.

Patients with occlusion myocardial infarction (OMI) and no ST-elevation on presenting electrocardiogram (ECG) are increasing in numbers. These patients have a poor prognosis and would benefit from immediate reperfusion therapy, but, currently, there are no accurate tools to identify them during initial triage. Here we report, to our knowledge, the first observational cohort study to develop machine learning models for the ECG diagnosis of OMI. Using 7,313 consecutive patients from multiple clinical sites, we derived and externally validated an intelligent model that outperformed practicing clinicians and other widely used commercial interpretation systems, substantially boosting both precision and sensitivity. Our derived OMI risk score provided enhanced rule-in and rule-out accuracy relevant to routine care, and, when combined with the clinical judgment of trained emergency personnel, it helped correctly reclassify one in three patients with chest pain. ECG features driving our models were validated by clinical experts, providing plausible mechanistic links to myocardial injury.

Effective Data Augmentation, Filters, and Automation Techniques for Automatic 12-Lead ECG Classification Using Deep Residual Neural Networks.

Automatic electrocardiogram (ECG) analysis plays a critical role in early detection and diagnosis of cardiac abnormalities and diseases. Data augmentation and automation strategies have been proposed to enhance the robustness of the machine and deep learning model for the classification of cardiac abnormalities. Here we propose 15 data augmentation and 6 filters, and an automation method using an end-to-end deep residual neural network (ResNet) model for automatic cardiac abnormalities detection from 12-lead ECG recordings. We evaluate the effectiveness of data augmentation/filtering and automation techniques using the proposed ResNet-based model on the China Physiological Signal Challenge (CPSC) dataset consisting of 9 diagnostic classes. The average F1 scores across 9 classes on the CPSC dataset trained with three data augmentation (baseline wander addition, dropout, and scaling) and a filter (sigmoid compression) were significantly higher than that without using augmentation/filters (baseline). The highest average F1 score with sigmoid compression method was significantly higher (relative improvement of 2.04 %) than the baseline while horizontal and vertical flipping augmentations were detrimental to the classification performance. Additionally, the results show that the random combination of four selected data augmentation and filter using the modified RandAugment technique provided a significantly higher average F1 score (relative improvement of 2.54 %) compared to the baseline. The proposed data augmentation, filters, and automation techniques provide an effective solution to improve the classification performance of the end-to-end deep learning model from ECG recordings without changing the model hyperparameters and structure.

Diagnostic performance of a new ECG algorithm for reducing false positive cases in patients suspected acute coronary syndrome.

BACKGROUND: Early and correct diagnosis of ST-segment elevation myocardial infarction (STEMI) is crucial for providing timely reperfusion therapy. Patients with ischemic symptoms presenting with ST-segment elevation on the electrocardiogram (ECG) are preferably transported directly to a catheterization laboratory (Cath-lab) for primary percutaneous coronary intervention (PPCI). However, the ECG often contains confounding factors making the STEMI diagnosis challenging leading to false positive Cath-lab activation. The objective of this study was to test the performance of a standard automated algorithm against an additional high specificity setting developed for reducing the false positive STEMI calls. METHODS: We included consecutive patients with an available digital prehospital ECG triaged directly to Cath-lab for acute coronary angiography between 2009 and 2012. An adjudicated discharge diagnosis of STEMI or no myocardial infarction (no-MI) was assigned for each patient. The new automatic algorithm contains a feature to reduce false positive STEMI interpretation. The STEMI performance with the standard setting (STD) and the high specificity setting (HiSpec) was tested against the adjudicated discharge diagnosis in a retrospective manner. RESULTS: In total, 2256 patients with an available digital prehospital ECG (mean age 63 ± 13 years, male gender 71%) were included in the analysis. The discharge diagnosis of STEMI was assigned in 1885 (84%) patients. The STD identified 165 true negative and 1457 true positive (206 false positive and 428 false negative) cases (77.3%, 44.5%, 87.6% and 17.3% for sensitivity, specificity, PPV and NPV, respectively). The HiSpec identified 191 true negative and 1316 true positive (180 false positive and 569 false negative) cases (69.8%, 51.5%, 88.0% and 25.1% for sensitivity, specificity, PPV and NPV, respectively). From STD to HiSpec, false positive cases were reduced by 26 (12,6%), but false negative results were increased by 33%. CONCLUSIONS: Implementing an automated ECG algorithm with a high specificity setting was able to reduce the number of false positive STEMI cases. However, the predictive values for both positive and negative STEMI identification were moderate in this highly selected STEMI population. Finally, due the reduced sensitivity/increased false negatives, a negative AMI statement should not be solely based on the automated ECG statement.

Performance and limitations of automated ECG interpretation statements in patients with suspected acute coronary syndrome.

BACKGROUND: The 12­lead ECG plays an important role in triaging patients with symptomatic coronary artery disease, making automated ECG interpretation statements of "Acute MI" or "Acute Ischemia" crucial, especially during prehospital transport when access to physician interpretation of the ECG is limited. However, it remains unknown how automated interpretation statements correspond to adjudicated clinical outcomes during hospitalization. We sought to evaluate the diagnostic performance of prehospital automated interpretation statements to four well-defined clinical outcomes of interest: confirmed ST- segment elevation myocardial infarction (STEMI); presence of actionable coronary culprit lesions, myocardial necrosis, or any acute coronary syndrome (ACS). METHODS: An observational cohort study that enrolled consecutive patients with non-traumatic chest pain transported via ambulance. Prehospital ECGs were obtained with the Philips MRX monitor from the medical command center and re-processed using manufacturer-specific diagnostic algorithms to denote the likelihood of >>>Acute MI<<< or >>>Acute Ischemia<<<. Two independent reviewers retrospectively adjudicated the study outcomes and disagreements were resolved by a third reviewer. RESULTS: Our study included 2400 patients (age 59 ± 16, 47% females, 41% Black), with 190 (8%) patients with documented automated diagnostic statements of acute MI or acute ischemia. The sensitivity/specificity of the automated algorithm for detecting confirmed STEMI (n = 143, 6%); presence of actionable coronary culprit lesions (n = 258, 11%), myocardial necrosis (n = 291, 12%), or any ACS (n = 378, 16%) were 62.9%/95.6%; 37.2%/95.6%; 38.5%/96.4%; and 30.7%/96.3%, respectively. CONCLUSION: Although being very specific, automated interpretation statements of acute MI/acute ischemia on prehospital ECGs are not satisfactorily sensitive to exclude symptomatic coronary disease. Patients without these automated interpretation statements should be considered further for significant underlying coronary disease based on the clinical context. TRIAL REGISTRATION: ClinicalTrials.gov # NCT04237688.

ECG reading differences demonstrated on two databases.

Many studies that rely on manual ECG interpretation as a reference use multiple ECG expert interpreters and a method to resolve differences between interpreters, reflecting the fact that experts sometimes use different criteria. The aim of this study was to show the effect of manual ECG interpretation style on training automated ECG interpretation. METHODS: The effect of ECG interpretation style or differing ECG criteria on algorithm training was shown in this study by careful analysis of the changes in algorithm performance when the algorithm was trained on one database and tested on a different database. Morphology related ECG interpretation was summarized in eleven abnormalities such as left bundle branch block (LBBB) and old anterior myocardial infarction (MI). Each of the two databases used in the study had a reference interpretation mapped to those eleven abnormalities. F1 algorithm performance scores across abnormalities were compared for four cases. First, the algorithm was trained and tested on randomly split database A and then trained on the training set of database A and tested on randomly chosen test set of database B. The previous two test cases were repeated for opposite databases, train and test on database B and then train on database B and test on the test set of database A. RESULTS: F1 scores across abnormalities were generally higher when training and testing on the same database. F1 scores were high for bundle branch blocks (BBB) no matter the training and testing database combination. Old anterior MI F1 score dropped for one cross-database comparison and not the other suggesting a difference in manual interpretation. CONCLUSION: For some abnormalities, human experts appear to have used different criteria for ECG interpretation, as evident by the difference between cross-database and within-database performance. Bundle branch blocks appear to be interpreted in a consistent manner.

Big data reveals insights for lead importance in ECG interpretation.

BACKGROUND: Not every lead contributes equally in the interpretation of an ECG. There are some abnormalities in which the lead importance is not clear either from cardiac electrophysiology or experience. Therefore, it is beneficial to develop an algorithm to quantify the lead importance in the reading of ECGs, namely to determine how much to weigh the evidence from each individual lead when interpreting ECG. METHODS: One representative beat per ECG lead was constructed for each ECG in a database. An algorithm was developed to find the top K (K = 1, 5, 10, 20, 50, 100) ECGs in the database that had the most similar morphology to the query ECG, independently for each lead. For each lead, the query ECG was interpreted based on the weighted average voting on the most similar ECGs by applying a variety of thresholds. For each category of abnormality, we found the threshold that maximized the median F1 score of sensitivity and positive predictive value among all ECG leads. Finally, the F1 score of each lead at this chosen threshold was defined as the importance value for that lead. RESULTS: Eighteen morphology-based categories of abnormality were investigated for two databases. For most, the lead importance confirmed what expert ECG readers already know. However, it also revealed new insights. For example, lead aVR appeared in the top 6 most important leads in 11 and 12 categories of abnormality in two databases respectively, and ranked first among 12 leads if summarizing all categories. CONCLUSIONS: Lead importance information may be useful in selecting only the most important leads to screen for a specific abnormality, for example using wearable patches.

Novel ECG features and machine learning to optimize culprit lesion detection in patients with suspected acute coronary syndrome.

BACKGROUND: Novel temporal-spatial features of the 12­lead ECG can conceptually optimize culprit lesions' detection beyond that of classical ST amplitude measurements. We sought to develop a data-driven approach for ECG feature selection to build a clinically relevant algorithm for real-time detection of culprit lesion. METHODS: This was a prospective observational cohort study of chest pain patients transported by emergency medical services to three tertiary care hospitals in the US. We obtained raw 10-s, 12­lead ECGs (500 s/s, HeartStart MRx, Philips Healthcare) during prehospital transport and followed patients 30 days after the encounter to adjudicate clinical outcomes. A total of 557 global and lead-specific features of P-QRS-T waveform were harvested from the representative average beats. We used Recursive Feature Elimination and LASSO to identify 35/557, 29/557, and 51/557 most recurrent and important features for LAD, LCX, and RCA culprits, respectively. Using the union of these features, we built a random forest classifier with 10-fold cross-validation to predict the presence or absence of culprit lesions. We compared this model to the performance of a rule-based commercial proprietary software (Philips DXL ECG Algorithm). RESULTS: Our sample included 2400 patients (age 59 ± 16, 47% female, 41% Black, 10.7% culprit lesions). The area under the ROC curves of our random forest classifier was 0.85 ± 0.03 with sensitivity, specificity, and negative predictive value of 71.1%, 84.7%, and 96.1%. This outperformed the accuracy of the automated interpretation software of 37.2%, 95.6%, and 92.7%, respectively, and corresponded to a net reclassification improvement index of 23.6%. Metrics of ST80; Tpeak-Tend; spatial angle between QRS and T vectors; PCA ratio of STT waveform; T axis; and QRS waveform characteristics played a significant role in this incremental gain in performance. CONCLUSIONS: Novel computational features of the 12­lead ECG can be used to build clinically relevant machine learning-based classifiers to detect culprit lesions, which has important clinical implications.

In Search of an Optimal Subset of ECG Features to Augment the Diagnosis of Acute Coronary Syndrome at the Emergency Department.

Background Classical ST-T waveform changes on standard 12-lead ECG have limited sensitivity in detecting acute coronary syndrome (ACS) in the emergency department. Numerous novel ECG features have been previously proposed to augment clinicians' decision during patient evaluation, yet their clinical utility remains unclear. Methods and Results This was an observational study of consecutive patients evaluated for suspected ACS (Cohort 1 n=745, age 59±17, 42% female, 15% ACS; Cohort 2 n=499, age 59±16, 49% female, 18% ACS). Out of 554 temporal-spatial ECG waveform features, we used domain knowledge to select a subset of 65 physiology-driven features that are mechanistically linked to myocardial ischemia and compared their performance to a subset of 229 data-driven features selected by multiple machine learning algorithms. We then used random forest to select a final subset of 73 most important ECG features that had both data- and physiology-driven basis to ACS prediction and compared their performance to clinical experts. On testing set, a regularized logistic regression classifier based on the 73 hybrid features yielded a stable model that outperformed clinical experts in predicting ACS, with 10% to 29% of cases reclassified correctly. Metrics of nondipolar electrical dispersion (ie, circumferential ischemia), ventricular activation time (ie, transmural conduction delays), QRS and T axes and angles (ie, global remodeling), and principal component analysis ratio of ECG waveforms (ie, regional heterogeneity) played an important role in the improved reclassification performance. Conclusions We identified a subset of novel ECG features predictive of ACS with a fully interpretable model highly adaptable to clinical decision support applications. Registration URL: https://www.clinicaltrials.gov; Unique Identifier: NCT04237688.

Detection of strict left bundle branch block by neural network and a method to test detection consistency.

OBJECTIVE: To develop an automatic algorithm to detect strict left bundle branch block (LBBB) on electrocardiograms (ECG) and propose a procedure to test the consistency of neural network detections. APPROACH: The database for the classification of strict LBBB was provided by Telemetric and Holter ECG Warehouse. It contained 10 s ECGs taken from the MADIT-CRT clinical trial. The database was divided into a training dataset (N = 300, strict LBBB = 174, non-strict LBBB = 126) and a test dataset (N = 302, strict LBBB = 156, non-strict LBBB = 146). LBBB-related features were extracted by Philips DXL™ algorithm, selected by a random forest classifier, and fed into a 5-layer neural network (NN) for the classification of strict LBBB on the training dataset. The performance of NN on the test dataset was compared to two random forest classifiers, an algorithm applying strict LBBB criteria, a wavelet-based approach, and a support-vector-machine approach. The consistency of NN's detection was tested on 549 2 min recordings of the PTB diagnostic ECG database. LBBB annotations are not required to measure consistency. MAIN RESULTS: The performance of NN on the test dataset were sensitivity = 91. 7%, specificity = 85.6% and accuracy = 88.7% (PPV = 87.2%, NPV = 90.6%). The consistency score of strict-LBBB and non-strict-LBBB detection was 0.9341 and 0.9973 respectively. CONCLUSION: NN achieved the highest specificity, accuracy, and PPV. Using random forest for feature selection and NN for classification increased interpretability and reduced computational cost. The consistency test showed that NN achieved high consistency scores in the detection of strict LBBB. SIGNIFICANCE: This work proposed an approach for selecting features and training NN for the detection of strict LBBB as well as a consistency test for black-box algorithms.

12-Lead ECG interpretation by database comparison.

BACKGROUND: Automated ECG interpretation is most often a rule-based expert system, though experts may disagree on the exact ECG criteria. One method to automate ECG analysis while indirectly using varied sets of expert rules is to base the automated interpretation on similar ECGs that already have a physician interpretation. The aim of this study is to develop and test an ECG interpretation algorithm based on such similar ECGs. METHODS: The study database consists of approximately 146,000 sequential 12-lead 10 s ECGs taken over the course of three years from a single hospital. All patient ECGs were included. Computer interpretation was corrected by physicians as part of standard care. The ECG algorithm developed here consisted of an ECG similarity search along with a method for estimating the interpretation from a small set of similar ECGs. A second level of differential diagnosis differentiated ECG categories with substantial similarity, such as LVH and LBBB. Interpretation performance was tested by ROC analysis including sensitivity (SE), specificity (SP), positive predictive value (PPV) and area under the ROC curve (AUC). RESULTS: LBBB was the category with the best ECG interpretation performance with an AUC of 0.981 while RBBB, LAFB and ventricular paced rhythm also had an AUC at 0.95 or above. AUC was 0.9 and above for the ischemic repolarization abnormality, LVH, old anterior MI, and early repolarization categories. All other morphology categories had an AUC over 0.8. CONCLUSION: ECG interpretation by analysis of ECG similarity provides adequate ECG interpretation performance on an unselected database using only strategies to weight the interpretation from those similar ECGs. Although this algorithm may not be ready to replace rule-based computer ECG analysis, it may be a useful adjunct recommender.

The algorithmic performance of J-Tpeak for drug safety clinical trial.

INTRODUCTION: The interval from J-point to T-wave peak (JTp) in ECG is a new biomarker able to identify drugs that prolong the QT interval but have different ion channel effects. If JTp is not prolonged, the prolonged QT may be associated with multi ion channel block that may have low torsade de pointes risk. From the automatic ECG measurement perspective, accurate and repeatable measurement of JTp involves different challenges than QT. We evaluated algorithm performance and JTp challenges using the Philips DXL diagnostic 12/16/18-lead algorithm. Measurement of JTp represents a different use model. Standard use of corrected QT interval is clinical risk assessment on patients with cardiac disease or suspicion of heart disease. Drug safety trials involve a very different population - young healthy subjects - who commonly have J-waves, notches and slurs. Drug effects include difficult and unusual morphology such as flat T-waves, gentle notches, and multiple T-wave peaks. METHODS: The JTp initiative study provided ECGs collected from 22 young subjects (11 males and females) in randomized testing of dofetilide, quinidine, ranolazine, verapamil and placebo. We compare the JTp intervals between DXL algorithm and the FDA published measurements. The lead wise, vector-magnitude (VM), root-mean-square (RMS) and principal-component-analysis (PCA) representative beats were used to measure JTp and QT intervals. We also implemented four different methods for T peak detection for comparison. RESULTS: We found that JTp measurements were closer to the reference for combined leads RMS and PCA than individual leads. Differences in J-point location led to part of the JTp measurement difference because of the high prevalence of J-waves, notches and slurs. Larger differences were noted for drug effect causing multiple distinct T-wave peaks (Tp). The automated algorithm chooses the later peak while the reference was the earlier peak. Choosing among different algorithmic strategies in T peak measurement results in the tradeoff between stability and the accurate detection of calcium or sodium channel block. CONCLUSION: Measurement of JTp has different challenges than QT measurement. JTp measurement accuracy improved with combined leads RMS and PCA over lead II or V5.

Intelligent use of advanced capabilities of diagnostic ECG algorithms in a monitoring environment.

A large number of ST-elevation notifications are generated by cardiac monitoring systems, but only a fraction of them is related to the critical condition known as ST-segment elevation myocardial infarction (STEMI) in which the blockage of coronary artery causes ST-segment elevation. Confounders such as acute pericarditis and benign early repolarization create electrocardiographic patterns mimicking STEMI but usually do not benefit from a real-time notification. A STEMI screening algorithm able to recognize those confounders utilizing capabilities of diagnostic ECG algorithms in variation analysis of ST segments helps to avoid triggering a non-actionable ST-elevation notification. However, diagnostic algorithms are generally designed to analyze short ECG snapshots collected in low-noise resting position and hence are susceptible to high levels of noise common in a monitoring environment. We developed a STEMI screening algorithm which performs a real-time signal quality evaluation on the ECG waveform to select the segments with quality high enough for subsequent analysis by a diagnostic ECG algorithm. The STEMI notifications generated by this multi-stage STEMI screening algorithm are significantly fewer than ST-elevation notifications generated by a continuous ST monitoring strategy.

Brain penetrant liver X receptor (LXR) modulators based on a 2,4,5,6-tetrahydropyrrolo[3,4-c]pyrazole core.

Liver X receptor (LXR) agonists have been reported to lower brain amyloid beta (Aß) and thus to have potential for the treatment of Alzheimer's disease. Structure and property based design led to the discovery of a series of orally bioavailable, brain penetrant LXR agonists. Oral administration of compound 18 to rats resulted in significant upregulation of the expression of the LXR target gene ABCA1 in brain tissue, but no significant effect on Aß levels was detected.

Discovery of a Novel, Orally Efficacious Liver X Receptor (LXR) ß Agonist.

This article describes the application of Contour to the design and discovery of a novel, potent, orally efficacious liver X receptor ß (LXRß) agonist (17). Contour technology is a structure-based drug design platform that generates molecules using a context perceptive growth algorithm guided by a contact sensitive scoring function. The growth engine uses binding site perception and programmable growth capability to create drug-like molecules by assembling fragments that naturally complement hydrophilic and hydrophobic features of the protein binding site. Starting with a crystal structure of LXRß and a docked 2-(methylsulfonyl)benzyl alcohol fragment (6), Contour was used to design agonists containing a piperazine core. Compound 17 binds to LXRß with high affinity and to LXRα to a lesser extent, and induces the expression of LXR target genes in vitro and in vivo. This molecule served as a starting point for further optimization and generation of a candidate which is currently in human clinical trials for treating atopic dermatitis.

Automated detection of ventricular pre-excitation in pediatric 12-lead ECG.

BACKGROUND: With increased interest in screening of young people for potential causes of sudden death, accurate automated detection of ventricular pre-excitation (VPE) or Wolff-Parkinson-White syndrome (WPW) in the pediatric resting ECG is important. Several recent studies have shown interobserver variability when reading screening ECGs and thus an accurate automated reading for this potential cause of sudden death is critical. We designed and tested an automated algorithm to detect pediatric VPE optimized for low prevalence. METHODS: Digital ECGs with 12 leads or 15 leads (12-lead plus V3R, V4R and V7) were selected from multiple hospitals and separated into a testing and training database. Inclusion criterion was age less than 16 years. The reference for algorithm detection of VPE was cardiologist annotation of VPE for each ECG. The training database (n=772) consisted of VPE ECGs (n=37), normal ECGs (n=492) and a high concentration of conduction defects, RBBB (n=232) and LBBB (n=11). The testing database was a random sample (n=763). All ECGs were analyzed with the Philips DXL ECG Analysis algorithm for basic waveform measurements. Additional ECG features specific to VPE, mainly delta wave scoring, were calculated from the basic measurements and the average beat. A classifier based on decision tree bootstrap aggregation (tree bagger) was trained in multiple steps to select the number of decision trees and the 10 best features. The classifier accuracy was measured on the test database. RESULTS: The new algorithm detected pediatric VPE with a sensitivity of 78%, a specificity of 99.9%, a positive predictive value of 88% and negative predictive value of 99.7%. CONCLUSION: This new algorithm for detection of pediatric VPE performs well with a reasonable positive and negative predictive value despite the low prevalence in the general population.

Identification of exercise-induced ischemia using QRS slopes.

In this work we studied a computer-aided approach using QRS slopes as unconventional ECG features to identify the exercise-induced ischemia during exercise stress testing and demonstrated that the performance is comparable to the experts' manual analysis using standard criteria involving ST-segment depression. We evaluated the performance of our algorithm using a database including 927 patients undergoing exercise stress tests and simultaneously collecting the ECG recordings and SPECT results. High resolution 12-lead ECG recordings were collected continuously throughout the rest, exercise, and recovery phases. Patients in the database were classified into three categories of moderate/severe ischemia, mild ischemia, and normal according to the differences in sum of the individual segment scores for the rest and stress SPECT images. Philips DXL 16-lead diagnostic algorithm was run on all 10-s segments of 12-lead ECG recordings for each patient to acquire the representative beats, ECG fiducial points from the representative beats, and other ECG parameters. The QRS slopes were extracted for each lead from the averaged representative beats and the leads with highest classification power were selected. We employed linear discriminant analysis and measured the performance using 10-fold cross-validation. Comparable performance of this method to the conventional ST-segment analysis exhibits the classification power of QRS slopes as unconventional ECG parameters contributing to improved identification of exercise-induced ischemia.

Agreement between cardiologists and fellows in interpretation of ischemic electrocardiographic changes in acute myocardial infarction.

BACKGROUND: Time from symptom onset may not be the best indicator for choosing reperfusion therapy for patients presenting with acute ST-elevation myocardial infarction (STEMI); consequently ECG-based methods have been developed. METHODS: This study evaluated the inter-observer agreement between experienced cardiologists and junior doctors in identifying the ECG findings of the pre-infarction syndrome (PIS) and evolving myocardial infarction (EMI). The ECGs of 353 STEMI patients were independently analyzed by two cardiologists, one fellow in cardiology, one fellow in internal medicine and a medical student. The last two were given a half-hour introduction of the PIS/EMI-algorithm. RESULTS: The inter-observer reliability between all the investigators was found to be good according to kappa statistics (κ 0.632-0.790) for the whole study population. When divided into different subgroups, the inter-observer agreements were from good to very good between the cardiologists and the fellow in cardiology (κ 0.652 -0.813) and from moderate to good (κ 0.464-0.784) between the fellow in internal medicine, medical student and the others. CONCLUSIONS: The PIS and EMI ECG patterns are reliably identified by experienced cardiologists and can be easily adopted by junior doctors.

Automatic detection of ECG cable interchange by analyzing both morphology and interlead relations.

BACKGROUND: ECG cable interchange can generate erroneous diagnoses. For algorithms detecting ECG cable interchange, high specificity is required to maintain a low total false positive rate because the prevalence of interchange is low. In this study, we propose and evaluate an improved algorithm for automatic detection and classification of ECG cable interchange. METHOD: The algorithm was developed by using both ECG morphology information and redundancy information. ECG morphology features included QRS-T and P-wave amplitude, frontal axis and clockwise vector loop rotation. The redundancy features were derived based on the EASI™ lead system transformation. The classification was implemented using linear support vector machine. The development database came from multiple sources including both normal subjects and cardiac patients. An independent database was used to test the algorithm performance. Common cable interchanges were simulated by swapping either limb cables or precordial cables. RESULTS: For the whole validation database, the overall sensitivity and specificity for detecting precordial cable interchange were 56.5% and 99.9%, and the sensitivity and specificity for detecting limb cable interchange (excluding left arm-left leg interchange) were 93.8% and 99.9%. Defining precordial cable interchange or limb cable interchange as a single positive event, the total false positive rate was 0.7%. When the algorithm was designed for higher sensitivity, the sensitivity for detecting precordial cable interchange increased to 74.6% and the total false positive rate increased to 2.7%, while the sensitivity for detecting limb cable interchange was maintained at 93.8%. The low total false positive rate was maintained at 0.6% for the more abnormal subset of the validation database including only hypertrophy and infarction patients. CONCLUSION: The proposed algorithm can detect and classify ECG cable interchanges with high specificity and low total false positive rate, at the cost of decreased sensitivity for certain precordial cable interchanges. The algorithm could also be configured for higher sensitivity for different applications where a lower specificity can be tolerated.

Detection of culprit coronary lesion location in pre-hospital 12-lead ECG.

BACKGROUND: Pre-hospital 12-lead ECG interpretation is important because pre-hospital activation of the coronary catheterization laboratory reduces ST-segment elevation myocardial infarction (STEMI) discovery-to-treatment time. In addition, some ECG features indicate higher risk in STEMI such as proximal left anterior descending (LAD) culprit lesion location. The challenging nature of the pre-hospital environment can lead to noisier ECGs which make automated STEMI detection difficult. We describe an automated system to classify lesion location as proximal LAD, LAD, right coronary artery (RCA) and left circumflex (LCx) and test the performance on pre-hospital 12-lead ECG. METHODS: The overall classifier was designed from three linked classifiers to separate LAD from non-LAD (RCA or LCx) in the first step, RCA from LCx in a second classifier and proximal from non-proximal LAD in the third classifier. The proximal LAD classifier was designed for high specificity because the output may be used in the decision to modify treatment. The LCx classifier was designed for high specificity because RCA is dominant in most people. The system was trained on a set of emergency department ECGs (n=181) and tested on a set of pre-hospital ECGs (n=80). Both sets were based on a sequential sample starting with symptoms suggesting acute coronary syndromes. Culprit lesion location was determined from coronary catheterization laboratory reports. Inclusion criteria included STEMI interpretation by computer and culprit lesion with 70% or more narrowing. Algorithm accuracy was measured on the test set by sensitivity (SE), specificity (SP), and positive predictive value (PPV). RESULTS: SE, SP and PPV were 50, 100 and 100% respectively for proximal LAD lesion location; 90, 100 and 100% for all LAD; 98, 72 and 78% for RCA; and 50, 98 and 90% for LCx. Specificity and PPV were high for proximal LAD, LAD and LCx. Specificity and PPV are not as high for RCA by design since the RCA-LCx tradeoff favors high specificity in LCx. CONCLUSION: Although our test database is not large, algorithm performance suggests culprit lesion location can be reliably determined from pre-hospital ECG. Further research is needed however to evaluate the impact of automated culprit lesion location on patient treatment and outcomes.

An electronic environment and contact direction sensitive scoring function for predicting affinities of protein-ligand complexes in Contour(®).

Contour(®) is a computational structure-based drug design technology that grows drug-like molecules by assembling context sensitive fragments in well-defined binding pockets. The grown molecules are scored by a novel empirical scoring function developed using high-resolution crystal structures of diverse classes of protein-ligand complexes and associated experimental binding affinities. An atomic model bearing features of the valence bond and VSEPR theories embodying their molecular electronic environment has been developed for non-covalent intermolecular interactions. On the basis of atomic hybridization and polarization states, each atom is modeled by features representing electron lone pairs, p-orbitals, and polar and non-polar hydrogens. A simple formal charge model was used to differentiate between polar and non-polar atoms. The interaction energy and the desolvation contribution of the protein-ligand association energy is computed as a linear sum of pair-wise interactions and desolvation terms. The pair-wise interaction energy captures short-range positive electrostatic interactions via hydrogen bonds, electrostatic repulsion of like charges, and non-bond contacts. The desolvation energy is estimated by calculating the energy required to desolvate interaction surfaces of the protein and the ligand in the complex. The scoring function predicts binding energies of a diverse set of protein-ligand complexes used for training with a correlation coefficient of 0.61. It also performs equally well in predicting association energies of a diverse validation set of protein-ligand complexes with a correlation coefficient of 0.57, which is equivalent to or better than 12 other scoring functions tested against this set including X-Score, GOLD, and DrugScore.