Artificial intelligence cloud platform improves arrhythmia detection from insertable cardiac monitors to 25 cardiac rhythm patterns through multi-label classification.

BACKGROUND: Electrocardiogram (ECG) is the gold standard for the diagnosis of cardiac arrhythmias and other heart diseases. Insertable cardiac monitors (ICMs) have been developed to continuously monitor cardiac activity over long periods of time and to detect 4 cardiac patterns (atrial tachyarrhythmias, ventricular tachycardia, bradycardia, and pause). However, interpretation of ECG or ICM subcutaneous ECG (sECG) is time-consuming for clinicians. Artificial intelligence (AI) classifies ECG and sECG with high accuracy in short times. OBJECTIVE: To demonstrate whether an AI algorithm can expand ICM arrhythmia recognition from 4 to many cardiac patterns. METHODS: We performed an exploratory retrospective study with sECG raw data coming from 20 patients wearing a Confirm Rx™ (Abbott, Sylmar, USA) ICM. The sECG data were recorded in standard conditions and then analyzed by AI (Willem™, IDOVEN, Madrid, Spain) and cardiologists, in parallel. RESULTS: In nineteen patients, ICMs recorded 2261 sECGs in an average follow-up of 23 months. Within these 2261 sECG episodes, AI identified 7882 events and classified them according to 25 different cardiac rhythm patterns with a pondered global accuracy of 88%. Global positive predictive value, sensitivity, and F1-score were 86.77%, 83.89%, and 85.52% respectively. AI was especially sensitive for bradycardias, pauses, rS complexes, premature atrial contractions, and inverted T waves, reducing the median time spent to classify each sECG compared to cardiologists. CONCLUSION: AI can process sECG raw data coming from ICMs without previous training, extending the performance of these devices and saving cardiologists' time in reviewing cardiac rhythm patterns detection.

Rational and design of ST-segment elevation not associated with acute cardiac necrosis (LESTONNAC). A prospective registry for validation of a deep learning system assisted by artificial intelligence.

BACKGROUND: Patients with chest pain and persistent ST segment elevation (STE) may not have acute coronary occlusions or serum troponin curves suggestive of acute necrosis. Our objective is the validation and cost-effectiveness analysis of a diagnostic model assisted by artificial intelligence (AI). METHODS: Prospective multicenter registry in two groups of patients with STE: I) coronary arteries without significant lesions and without serum troponin curve suggestive of acute necrosis, II) myocardial infarction with acute coronary occlusion. The inclusion criteria are the following: 1) age ≥ 18 years, 2) chest pain or symptoms suggestive of myocardial ischemia, 3) STE at point J in two contiguous leads ≥0.1 mV, in V2 and V3 ≥ 0,2 mV and 4) signature of informed consent. The exclusion criteria are the following: 1) left bundle branch block, 2) acute cardiac necrosis in the absence of significant epicardial coronary artery stenosis, 3) STE ≤ 0.1 mV with pathologic Q wave, 4) severe anemia (hemoglobin <8.0 g/dl). For each patient without acute cardiac necrosis, the next patient from that center of the same sex and similar age (± 5 years) with myocardial infarction and acute coronary occlusion will be included. A manual centralized electrocardiographic analysis and another by deep learning AI will be performed. CONCLUSIONS: The results of the study will provide new information for the stratification of patients with STE. Our hypothesis is that an AI analysis of the surface electrocardiogram allows a better distinction of patients with STE due to acute myocardial ischemia, from those with another etiology.

Personalized monitoring of electrical remodelling during atrial fibrillation progression via remote transmissions from implantable devices.

AIMS: Atrial electrical remodelling (AER) is a transitional period associated with the progression and long-term maintenance of atrial fibrillation (AF). We aimed to study the progression of AER in individual patients with implantable devices and AF episodes. METHODS AND RESULTS: Observational multicentre study (51 centres) including 4618 patients with implantable cardioverter-defibrillator +/-resynchronization therapy (ICD/CRT-D) and 352 patients (2 centres) with pacemakers (median follow-up: 3.4 years). Atrial activation rate (AAR) was quantified as the frequency of the dominant peak in the signal spectrum of AF episodes with atrial bipolar electrograms. Patients with complete progression of AER, from paroxysmal AF episodes to electrically remodelled persistent AF, were used to depict patient-specific AER slopes. A total of 34 712 AF tracings from 830 patients (87 with pacemakers) were suitable for the study. Complete progression of AER was documented in 216 patients (16 with pacemakers). Patients with persistent AF after completion of AER showed ∼30% faster AAR than patients with paroxysmal AF. The slope of AAR changes during AF progression revealed patient-specific patterns that correlated with the time-to-completion of AER (R2 = 0.85). Pacemaker patients were older than patients with ICD/CRT-Ds (78.3 vs. 67.2 year olds, respectively, P < 0.001) and had a shorter median time-to-completion of AER (24.9 vs. 93.5 days, respectively, P = 0.016). Remote transmissions in patients with ICD/CRT-D devices enabled the estimation of the time-to-completion of AER using the predicted slope of AAR changes from initiation to completion of electrical remodelling (R2 = 0.45). CONCLUSION: The AF progression shows patient-specific patterns of AER, which can be estimated using available remote-monitoring technology.

Artificial intelligence augments detection accuracy of cardiac insertable cardiac monitors: Results from a pilot prospective observational study.

Background: Insertable cardiac monitors (ICMs) are indicated for long-term monitoring of patients with unexplained syncope or who are at risk for cardiac arrhythmias. The volume of ICM-transmitted information may result in long data review times to identify true and clinically relevant arrhythmias. Objective: The purpose of this study was to evaluate whether artificial intelligence (AI) may improve ICM detection accuracy. Methods: We performed a retrospective analysis of consecutive patients implanted with the Confirm RxTM ICM (Abbott) and followed in a prospective observational study. This device continuously monitors subcutaneous electrocardiograms (SECGs) and transmits to clinicians information about detected arrhythmias and patient-activated symptomatic episodes. All SECGs were classified by expert electrophysiologists and by the WillemTM AI algorithm (IDOVEN). Results: During mean follow-up of 23 months, of 20 ICM patients (mean age 68 ± 12 years; 50% women), 19 had 2261 SECGs recordings associated with cardiac arrhythmia detections or patient symptoms. True arrhythmias occurred in 11 patients: asystoles in 2, bradycardias in 3, ventricular tachycardias in 4, and atrial tachyarrhythmias (atrial tachycardia/atrial fibrillation [AT/AF]) in 10; with 6 patients having >1 arrhythmia type. AI algorithm overall accuracy for arrhythmia classification was 95.4%, with 97.19% sensitivity, 94.52% specificity, 89.74% positive predictive value, and 98.55% negative predictive value. Application of AI would have reduced the number of false-positive results by 98.0% overall: 94.0% for AT/AF, 87.5% for ventricular tachycardia, 99.5% for bradycardia, and 98.8% for asystole. Conclusion: Application of AI to ICM-detected episodes is associated with high classification accuracy and may significantly reduce health care staff workload by triaging ICM data.

Early prognostic value of an Algorithm based on spectral Variables of Ventricular fibrillAtion from the EKG of patients with suddEn cardiac death: A multicentre observational study (AWAKE).

OBJECTIVE: Ventricular fibrillation (VF)-related sudden cardiac death (SCD) is a leading cause of mortality and morbidity. Current biological and imaging parameters show significant limitations on predicting cerebral performance at hospital admission. The AWAKE study (NCT03248557) is a multicentre observational study to validate a model based on spectral ECG analysis to early predict cerebral performance and survival in resuscitated comatose survivors. METHODS: Data from VF ECG tracings of patients resuscitated from SCD will be collected using an electronic Case Report Form. Patients can be either comatose (Glasgow Coma Scale - GCS - ≤8) survivors undergoing temperature control after return of spontaneous circulation (RoSC), or those who regain consciousness (GCS=15) after RoSC; all admitted to Intensive Cardiac Care Units in 4 major university hospitals. VF tracings prior to the first direct current shock will be digitized and analyzed to derive spectral data and feed a predictive model to estimate favorable neurological performance (FNP). The results of the model will be compared to the actual prognosis. RESULTS: The primary clinical outcome is FNP during hospitalization. Patients will be categorized into 4 subsets of neurological prognosis according to the risk score obtained from the predictive model. The secondary clinical outcomes are survival to hospital discharge, and FNP and survival after 6 months of follow-up. The model-derived categorisation will be also compared with clinical variables to assess model sensitivity, specificity, and accuracy. CONCLUSIONS: A model based on spectral analysis of VF tracings is a promising tool to obtain early prognostic data after SCD.

QRS duration reflects underlying changes in conduction velocity during increased intraventricular pressure and heart failure.

Pressure overload and heart failure electrophysiological remodeling (HF-ER) in pigs are associated with decreased conduction velocity (CV) and dispersion of repolarization, which lead to higher risk of ventricular arrhythmia. This work aimed to establish the correlation between QRS complex duration and underlying changes in CV during increased intraventricular pressure (IVP) and/or HF-ER ex-vivo, and to determine whether QRS duration could be sensitive to an acute increase in left ventricular (LV) afterload in-vivo. HF-ER was induced in 7 pigs by high-rate ventricular pacing. Seven weight-matched animals were used as controls. Isolated Langendorff-perfused hearts underwent programmed ventricular stimulation to study QRS complex duration and CV under low/high IVP, using volume-conducted ECG and epicardial optical mapping, respectively. Four additional pigs underwent open-chest surgery to increase LV afterload by partially clamping the ascending aorta, while measuring QRS complex duration during sinus rhythm (SR). In 13 hearts included for analysis, both HF-ER and increased IVP showed significantly slower epicardial CV (-40% and -15%, p < 0.001 and p = 0.004, respectively), which correlated with similar widening of the QRS complex (+41% and +17%, p = 0.005 and p < 0.001, respectively). HF-ER hearts shower larger prolongation of the QRS complex than controls upon increasing the IVP (+21% vs. +12%, respectively. HF-ER*IVP interaction: p = 0.004). QRS complex widened after increasing LV afterload in-vivo (n=3), with correlation between QRS duration and aortic diastolic pressures (R = 0.58, p < 0.001). In conclusion, high IVP and/or HF-ER significantly decrease CV, which correlates with QRS widening on the ECG during ventricular pacing. Increased myocardial wall stress also widens the QRS complex during SR in-vivo.

Early prognostic value of an Algorithm based on spectral Variables of Ventricular fibrillAtion from the EKG of patients with suddEn cardiac death: A multicentre observational study (AWAKE) / Valor pronóstico precoz de un algoritmo basado en variables espectrales de la fibrilación ventricular del ECG de pacientes con muerte súbita cardiaca: estudio observacional multicéntrico (AWAKE)

Abstract Objective: Ventricular fibrillation (VF)-related sudden cardiac death (SCD) is a leading cause of mortality and morbidity. Current biological and imaging parameters show significant limitations on predicting cerebral performance at hospital admission. The AWAKE study (NCT03248557) is a multicentre observational study to validate a model based on spectral ECG analysis to early predict cerebral performance and survival in resuscitated comatose survivors. Methods: Data from VF ECG tracings of patients resuscitated from SCD will be collected using an electronic Case Report Form. Patients can be either comatose (Glasgow Coma Scale GCS --- ≤8) survivors undergoing temperature control after return of spontaneous circulation (RoSC), or those who regain consciousness (GCS = 15) after RoSC; all admitted to Intensive Cardiac Care Units in 4 major university hospitals. VF tracings prior to the first direct current shock will be digitized and analyzed to derive spectral data and feed a predictive model to estimate favorable neurological performance (FNP). The results of the model will be compared to the actual prognosis. Results: The primary clinical outcome is FNP during hospitalization. Patients will be categorized into 4 subsets of neurological prognosis according to the risk score obtained from the predictive model. The secondary clinical outcomes are survival to hospital discharge, and FNP and survival after 6 months of follow-up. The model-derived categorisation will be also compared with clinical variables to assess model sensitivity, specificity, and accuracy. Conclusions: A model based on spectral analysis of VF tracings is a promising tool to obtain early prognostic data after SCD.

Resumen Objetivo: La muerte súbita (MS) por fibrilación ventricular (FV) es una importante causa de morbilidad y mortalidad. Los métodos biológicos y de imagen actuales muestran limitaciones para predecir el pronóstico cerebral al ingreso hospitalario. AWAKE es un estudio observacional, multicéntrico, con el objetivo de validar un modelo basado en el análisis espectral del elec- trocardiograma (ECG), que predice precozmente el pronóstico cerebral y la supervivencia en pacientes resucitados y en estado de coma. Métodos: Se recogerán datos de los ECG con FV de pacientes reanimados de MS. Los pacientes pueden ser tanto supervivientes en estado de coma (Glasgow Coma Scale GCS ≤ 8) sometidos a control de temperatura tras la recuperación de circulación espontánea (RCE), como aquellos que recuperan la consciencia (GCS = 15) tras RCE; todos ellos ingresados en unidades de terapia intensiva cardiológica de 4 hospitales de referencia. Los registros de FV previos al primer choque se digitalizarán y analizarán para obtener datos espectrales que se incluirán en un modelo predictivo que estime el pronóstico neurológico favorable (PNF). El resultado del modelo se comparará con el pronóstico real. Resultados: El objetivo principal es el PNF durante la hospitalización. Los pacientes se categorizarán en 4 subgrupos de pronóstico neurológico según la estimación de riesgo obtenida en el modelo predictivo. Los objetivos secundarios son supervivencia al alta hospitalaria, y PNF y supervivencia a los 6 meses. El resultado de este modelo también se comparará con el pronóstico según variables clínicas. Conclusiones: Un modelo basado en el análisis espectral de registros de FV es una herramienta prometedora para obtener datos pronósticos precoces tras MS por FV.