Prediction Of Incident Atrial Fibrillation Research Articles

Abstract 4135754: Prediction of Incident Atrial Fibrillation Using Handheld Single-Lead Electrocardiograms from the VITAL-AF Trial

Background: Deep learning of 12-lead electrocardiograms (ECGs) can estimate future atrial fibrillation (AF) risk. Whether these models can be applied to single-lead ECGs (1L ECGs) from wearable or handheld devices is unknown. Aim: To apply a previously published deep learning model to estimate 2-year incident AF risk using handheld 1L ECGs obtained prospectively in a large AF screening trial. Methods: In the VITAL-AF trial (NCT03515057), over 16,000 primary care patients aged ≥65 years underwent AF screening with handheld 1L ECG. We applied our published deep learning model (ECG-AI) derived using single leads of over 450,000 standard 12-lead ECGs external to the VITAL-AF study (1L ECG-AI). We compared the performance of 1L ECG-AI to the validated CHARGE-AF clinical risk score and calculated net reclassification indices. Results: Among 15,694 individuals enrolled in VITAL-AF and without prevalent AF (age 74±7 years, 9,104 [58%] women), 1L ECG-AI discriminated 2-year incident AF (area under receiver operating characteristic curve [AUROC] 0.666 [95% CI 0.603-0.721]). With addition of age and sex (1L ECG-AI AS), 2-year AF discrimination (0.695 [0.637-0.742]) was comparable to the 11-component CHARGE-AF clinical risk score (0.679 [0.625-0.729]). Two-year AF incidence was markedly higher with 1L ECG-AI AS in the top 5% (8.0% [5.7-10.2]) vs bottom 5% (0.89% [0.23-0.55]). At a threshold of ≥3% estimated 2-year AF risk, AF incidence was progressively higher among patients at high risk according to neither model (1.5% [1.2-1.8]), one model (1L ECG-AI: 3.3% [2.3-4.2]; CHARGE-AF 3.4% [2.2-4.6]), and both models (5.8% [5.0-6.5]), implying that clinical risk and ECG-AI signals are complementary ( Figure ). Compared to screening all individuals at the guideline-based threshold of ≥65 years, 1L ECG-AI AS resulted in favorable net reclassification improvement (0.27 [0.22-0.32]). Conclusion: An ECG-AI algorithm developed using single leads of a 12-lead ECG combined with age and sex can discriminate AF risk using real-world handheld 1L ECGs with comparable performance to the CHARGE-AF score. 1L ECG-AI signals complement clinical risk factors. ECG-AI applied to 1L ECG may increase the reach and efficiency of AF screening.

Incident atrial fibrillation prediction using ECG-based deep learning at a specialized tertiary cardiac care center

Abstract Background Deep learning (DL) applied to electrocardiograms (ECG) is an emerging modality in the prediction of incident atrial fibrillation or atrial flutter (AF). The generalizability of such methods to a tertiary heart center (THC) population has not yet been formally explored. Purpose Evaluate the performance of DL in predicting 5-year incident AF using a resting 12-lead ECG in sinus rhythm acquired at a THC. Methods In a retrospective study, we examined 1.4 million ECGs acquired from over 250,000 adults at a THC between 2004 and 2022. ECGs were excluded if they were performed within 30 days of cardiac surgery, captured a rhythm other than sinus rhythm, or were acquired in patients with pre-existing AF. Incident AF at 5 years was modelled as a binary outcome and was determined on the basis of available outpatient and inpatient clinical databases and ECG diagnoses. Included ECGs were randomly split by distributing patients into training (70%), validation (10%), and test (20%) sets. A ResNet-50 model was trained using the training set. Hyperparameters were optimized using the validation set. The tuned model's performance is reported on the test set. The results at the patient level were derived by averaging the model's probability outputs for ECGs grouped according to both their AF outcome and the patient's identity. Bootstrapping was used to report confidence intervals (CI). Sensitivity and specificity are reported at a classification threshold based on the Matthews correlation coefficient. Saliency maps were used to enhance the model’s explainability. Results A total of 669,782 ECGs (47% of the screened ECGs) were included among 145,323 patients. Mean age was 63±15 years and 62% were male. The 5-year incident AF outcome was observed in 12% of ECGs and 16% of patients. The performance of the tuned model was first evaluated at the ECG level on the test set demonstrating an area under the receiver operating curve (AUC) of 0.75 (95% CI: 0.745-0.753) and an integrated calibration index of 0.009 (95% CI: 0.008-0.011). When testing the model at the patient level to simulate a deployment scenario, the AUC improved to 0.78 (95% CI: 0.768-0.783) with a sensitivity of 50% (95% CI: 49-52) and specificity of 87% (95% CI: 86.5-87.3). Stratified results by sex showed a better AUC in females, 0.81 (95% CI: 0.80-0.82), compared to males, 0.75 (95% CI: 0.74-0.76). Subgroup analyses revealed a trend of improved performance at the patient level with an increasing number of ECGs. Saliency maps highlighted the P-wave area as having the highest influence on the model’s prediction, thereby enhancing the model’s plausibility and explainability. Conclusion A unimodal ECG-based deep learning model showed promising 5-year incident AF prediction performance in a cohort of all-comer patients at a tertiary heart center. Further studies could explore the use of multimodal prediction models that integrate ECGs with other clinical and imaging data.

Prediction of incident atrial fibrillation using deep learning, clinical models and polygenic scores.

Deep learning applied to electrocardiograms (ECG-AI) is an emerging approach for predicting atrial fibrillation or flutter (AF). This study introduces an ECG-AI model developed and tested at a tertiary cardiac centre, comparing its performance with clinical and AF polygenic scores (PGS). ECG in sinus rhythm from the Montreal Heart Institute were analysed, excluding those from patients with preexisting AF. The primary outcome was incident AF at 5 years. An ECG-AI model was developed by splitting patients into non-overlapping datasets: 70% for training, 10% for validation, and 20% for testing. Performance of ECG-AI, clinical models and PGS was assessed in the test dataset. The ECG-AI model was externally validated in the Medical Information Mart for Intensive Care-IV (MIMIC-IV) hospital dataset. A total of 669,782 ECGs from 145,323 patients were included. Mean age was 61±15 years, and 58% were male. The primary outcome was observed in 15% of patients and the ECG-AI model showed an area under the receiver operating characteristic curve (AUC) of 0.78. In time-to-event analysis including the first ECG, ECG-AI inference of high risk identified 26% of the population with a 4.3-fold increased risk of incident AF (95% confidence interval 4.02-4.57). In a subgroup analysis of 2,301 patients, ECG-AI outperformed CHARGE-AF (AUC=0.62) and PGS (AUC=0.59). Adding PGS and CHARGE-AF to ECG-AI improved goodness-of-fit (likelihood ratio test p<0.001), with minimal changes to the AUC (0.76-0.77). In the external validation cohort (mean age 59±18 years, 47% male, median follow-up 1.1 year) ECG-AI model performance= remained consistent (AUC=0.77). ECG-AI provides an accurate tool to predict new-onset AF in a tertiary cardiac centre, surpassing clinical and polygenic scores.

Innovative approaches to atrial fibrillation prediction: should polygenic scores and machine learning be implemented in clinical practice?

Atrial fibrillation (AF) prediction and screening are of important clinical interest because of the potential to prevent serious adverse events. Devices capable of detecting short episodes of arrhythmia are now widely available. Although it has recently been suggested that some high-risk patients with AF detected on implantable devices may benefit from anticoagulation, long-term management remains challenging in lower-risk patients and in those with AF detected on monitors or wearable devices as the development of clinically meaningful arrhythmia burden in this group remains unknown. Identification and prediction of clinically relevant AF is therefore of unprecedented importance to the cardiologic community. Family history and underlying genetic markers are important risk factors for AF. Recent studies suggest a good predictive ability of polygenic risk scores, with a possible additive value to clinical AF prediction scores. Artificial intelligence, enabled by the exponentially increasing computing power and digital data sets, has gained traction in the past decade and is of increasing interest in AF prediction using a single or multiple lead sinus rhythm electrocardiogram. Integrating these novel approaches could help predict AF substrate severity, thereby potentially improving the effectiveness of AF screening and personalizing the management of patients presenting with conditions such as embolic stroke of undetermined source or subclinical AF. This review presents current evidence surrounding deep learning and polygenic risk scores in the prediction of incident AF and provides a futuristic outlook on possible ways of implementing these modalities into clinical practice, while considering current limitations and required areas of improvement.

P-wave characteristics as electrocardiographic markers of atrial abnormality in prediction of incident atrial fibrillation – The Malmö Preventive Project

BackgroundP-wave indices reflect atrial abnormalities contributing to atrial fibrillation (AF). We aimed to assess a comprehensive set of P-wave characteristics for prediction of incident AF in a population-based setting. MethodsMalmö Preventative Project (MPP) participants were reexamined in 2002–2006 with electrocardiographic (ECG) and echocardiographic examinations and followed for 5 years. AF-free subjects (n = 983, age 70 ± 5 years, 38% females) with sinus rhythm ECGs were included in the study. ECGs were digitally processed using the Glasgow algorithm. P-wave duration, axis, dispersion, P-terminal force in lead V1 and interatrial block (IAB) were evaluated. ECG risk score combining the morphology, voltage and length of P-wave (MVP score) was calculated. New-onset diagnoses of AF were obtained from nation-wide registers. ResultsDuring follow up, 66 patients (7%) developed AF. After adjustment for age and gender, the independent predictors of AF were abnormal P-wave axis > 75° (HR 1.63 CI95% 1.95–11.03) and MVP score 4 (HR 6.17 CI 95% 1.76–21.64), both correlated with LA area: Person r − 0.146, p < 0.001 and 0.192, p < 0.001 respectively. Advanced IAB (aIAB) with biphasic P-wave morphology in leads III and aVF was the most prevalent variant of aIAB and predicted AF in a univariate model (HR 2.59 CI 95% 1.02–6.58). ConclusionP-wave frontal axis and MVP score are ECG-based AF predictors in the population-based cohort. Our study provides estimates for prevalence and prognostic importance of different variants of aIAB, providing a support to use biphasic P-wave morphology in lead aVF as the basis for aIAB definition.

Prediction of incident atrial fibrillation in hypertrophic cardiomyopathy

Background and aimAtrial fibrillation (AF) is the most common sustained arrhythmia in hypertrophic cardiomyopathy (HCM) with significant effects on outcome. We aim to compare the left atrial (LA) diameter measurement with HCM-AF Score in predicting atrial fibrillation (AF) development in HCM. MethodsFrom the regional cohort of the Campania Region, Italy, 519 HCM patients (38% women, age45 ± 17 years) without history of AF, were enrolled in the study. The primary clinical endpoint was the development of AF, defined as at least 1 episode documented by ECG. ResultsDuring the follow-up (mean 8 ± 6, IQ range 2.5–11.2 years), 99 patients (19%) developed AF. Patients who developed AF were more symptomatic, had higher prevalence of ICD implantation, had larger LA diameter, greater left ventricular (LV) maximal wall thickness and LV outflow tract obstruction (p < 0.01). Both LA diameter and HCM-AF score were higher in patients who developed AF versus those who did not (LA diameter 49 ± 7 versus 43 ± 6 mm; HCM-AF score 22 ± 4 versus 19 ± 4; p < 0.0001); however, ROC curve analysis demonstrated that LA diameter had a significant greater area under the curve than HCM-AF Score (p < 0.0001). At 5 years follow-up, a LA diameter > 46 mm, showed a similar accuracy in predicting AF development of HCM-AF score ≥ 22, which identifies patients at high risk to develop AF. ConclusionOur analysis shows that LA diameter, a worldwide and simple echocardiographic measure, is capable alone to predict AF development in HCM patients.

Relative fat mass and prediction of incident atrial fibrillation, heart failure and coronary artery disease in the general population.

Relative fat mass (RFM) is an emerging marker of obesity that estimates body fat percentage using a sex-specific formula containing height and waist circumference (WC). We studied the association of RFM with incident atrial fibrillation (AF), heart failure (HF), and coronary artery disease (CAD) and explored RFM cutoffs for cardiovascular disease (CVD) prediction. We studied 95,003 participants (age 45 ± 13 years, 59% women) without prevalent AF, HF or CAD from the population-based Lifelines study. Outcomes were ascertained using electrocardiography and self-reported questionnaire data. We used logistic regression to study the association of RFM with individual outcomes and a composite outcome (incident AF, HF, and/or CAD). Multivariable models were adjusted for components of the SCORE risk model (age, sex, systolic blood pressure, cholesterol, and smoking). Optimal cutoffs were determined using the Youden index. During a median follow-up of 3.8 (3.0-4.6) years, 224 (0.2%) participants developed AF, 1003 (1.1%) HF and 657 (0.7%) CAD. After multivariable adjustment, RFM was significantly associated with all outcomes (standardised OR 1.26, 95% CI 1.18-1.34 for the composite outcome). Optimal RFM cutoffs ( ≥26 for men, ≥38 for women) were lower than previously proposed RFM cutoffs ( ≥30 for men, ≥40 for women). In general, overall discriminative ability of RFM and its cutoffs was at least similar (in women) or better (in men) compared to BMI and WC. Since RFM was substantially correlated with age, we additionally determined age-specific cutoffs, which ranged from 23 to 27 in men and 33 to 43 in women. RFM is associated with incident AF, HF, and CAD and may be used as a simple and intuitive marker of obesity and cardiovascular risk in the general population. This study provides potential RFM cutoffs for CVD prediction that may be used by future studies or preventive strategies targeting obesity and cardiovascular risk.

Inflammatory proteomics profiling for prediction of incident atrial fibrillation

ObjectiveAtrial fibrillation (AF) has emerged as a common condition in older adults. Cardiovascular risk factors only explain about 50% of AF cases. Inflammatory biomarkers may help close this gap as...

Exploring the incremental utility of circulating biomarkers for robust risk prediction of incident atrial fibrillation in European cohorts using regressions and modern machine learning methods.

To identify robust circulating predictors for incident atrial fibrillation (AF) using classical regressions and machine learning (ML) techniques within a broad spectrum of candidate variables. In pooled European community cohorts (n = 42 280 individuals), 14 routinely available biomarkers mirroring distinct pathophysiological pathways including lipids, inflammation, renal, and myocardium-specific markers (N-terminal pro B-type natriuretic peptide [NT-proBNP], high-sensitivity troponin I [hsTnI]) were examined in relation to incident AF using Cox regressions and distinct ML methods. Of 42 280 individuals (21 843 women [51.7%]; median [interquartile range, IQR] age, 52.2 [42.7, 62.0] years), 1496 (3.5%) developed AF during a median follow-up time of 5.7 years. In multivariable-adjusted Cox-regression analysis, NT-proBNP was the strongest circulating predictor of incident AF [hazard ratio (HR) per standard deviation (SD), 1.93 (95% CI, 1.82-2.04); P < 0.001]. Further, hsTnI [HR per SD, 1.18 (95% CI, 1.13-1.22); P < 0.001], cystatin C [HR per SD, 1.16 (95% CI, 1.10-1.23); P < 0.001], and C-reactive protein [HR per SD, 1.08 (95% CI, 1.02-1.14); P = 0.012] correlated positively with incident AF. Applying various ML techniques, a high inter-method consistency of selected candidate variables was observed. NT-proBNP was identified as the blood-based marker with the highest predictive value for incident AF. Relevant clinical predictors were age, the use of antihypertensive medication, and body mass index. Using different variable selection procedures including ML methods, NT-proBNP consistently remained the strongest blood-based predictor of incident AF and ranked before classical cardiovascular risk factors. The clinical benefit of these findings for identifying at-risk individuals for targeted AF screening needs to be elucidated and tested prospectively.

Prediction of incident atrial fibrillation in post-stroke patients using machine learning: A French nationwide study

Prediction of incident atrial fibrillation in post-stroke patients using machine learning: A French nationwide study

Prediction of incident atrial fibrillation in post-stroke patients using machine learning: a French nationwide study.

Targeting ischemic strokes patients at risk of incident atrial fibrillation (AF) for prolonged cardiac monitoring and oral anticoagulation remains a challenge. Clinical risk scores have been developed to predict post-stroke AF with suboptimal performances. Machine learning (ML) models are developing in the field of AF prediction and may be used to discriminate post-stroke patients at risk of new onset AF. This study aimed to evaluate ML models for the prediction of AF and to compare predictive ability to usual clinical scores. Based on a French nationwide cohort of 240,459 ischemic stroke patients without AF at baseline from 2009 to 2012, ML models were trained on a train set and the best model was selected to be evaluate on the test set. Discrimination of the best model was evaluated using the C index. We finally compared our best model with previously described clinical scores. During a mean follow-up of 7.9 ± 11.5months, 14,095 patients (mean age 77.6 ± 10.6; 50.3% female) developed incident AF. After training, the best ML model selected was a deep neural network with a C index of 0.77 (95% CI 0.76-0.78) on the test set. Compared to traditional clinical scores, the selected model was statistically significantly superior to the CHA2DS2-VASc score, Framingham risk score, HAVOC score and C2HEST score (P < 0.0001). The ability to predict AF was improved as shown by net reclassification index increase (P < 0.0001) and decision curve analysis. ML algorithms predict incident AF post-stroke with a better ability than previously developed clinical scores. AF: atrial fibrillation; DNN: deep neural network; IS: ischemic stroke; KNN: K-nearest neighbors; LR: logistic regression; RFC: random forest classifier; XGBoost: extreme gradient boosting.

Prediction of incident cardiovascular events using machine learning and CMR radiomics

ObjectivesEvaluation of the feasibility of using cardiovascular magnetic resonance (CMR) radiomics in the prediction of incident atrial fibrillation (AF), heart failure (HF), myocardial infarction (MI), and stroke using machine learning techniques.MethodsWe identified participants from the UK Biobank who experienced incident AF, HF, MI, or stroke during the continuous longitudinal follow-up. The CMR indices and the vascular risk factors (VRFs) as well as the CMR images were obtained for each participant. Three-segmented regions of interest (ROIs) were computed: right ventricle cavity, left ventricle (LV) cavity, and LV myocardium in end-systole and end-diastole phases. Radiomics features were extracted from the 3D volumes of the ROIs. Seven integrative models were built for each incident cardiovascular disease (CVD) as an outcome. Each model was built with VRF, CMR indices, and radiomics features and a combination of them. Support vector machine was used for classification. To assess the model performance, the accuracy, sensitivity, specificity, and AUC were reported.ResultsAF prediction model using the VRF+CMR+Rad model (accuracy: 0.71, AUC 0.76) obtained the best result. However, the AUC was similar to the VRF+Rad model. HF showed the most significant improvement with the inclusion of CMR metrics (VRF+CMR+Rad: 0.79, AUC 0.84). Moreover, adding only the radiomics features to the VRF reached an almost similarly good performance (VRF+Rad: accuracy 0.77, AUC 0.83). Prediction models looking into incident MI and stroke reached slightly smaller improvement.ConclusionsRadiomics features may provide incremental predictive value over VRF and CMR indices in the prediction of incident CVDs.Key Points• Prediction of incident atrial fibrillation, heart failure, stroke, and myocardial infarction using machine learning techniques.• CMR radiomics, vascular risk factors, and standard CMR indices will be considered in the machine learning models.• The experiments show that radiomics features can provide incremental predictive value over VRF and CMR indices in the prediction of incident cardiovascular diseases.

The predictive value of abnormal P-wave axis for the detection of incident atrial fibrillation: A systematic review with meta-analysis.

There is growing interest in the prediction of incident atrial fibrillation (AF). The 12-lead electrocardiogram (ECG) has been a particularly rich target for possible prediction strategies. The P-wave axis is an ECG parameter that reflects the dominant vector of atrial depolarisation and is usually 0° -75°. There is a large body of literature suggesting that AF reflects structural and conduction abnormalities of the atria, and thus the P-wave axis may represent a sensitive parameter to detect such changes. A systematic review and meta-analysis of published literature associating abnormal P-wave axis and the development of incident AF was performed. Electronic databases were systematically searched from inception to October 2021. A random-effects model with generic inverse variance weights was utilised to pool the most adjusted effect measure from each paper. A funnel plot was used to assess publication bias. After excluding duplicate studies, 568 studies were screened. A total of eleven studies were identified that associated an abnormal P-wave axis with the subsequent detection of AF. The eight studies that considered abnormal P-wave axis as being <0° or >75° were pooled for meta-analysis. In the pooled studies a total of 78,222 patients were included with 5656 cases of incident atrial fibrillation identified. The meta-analysis of the studies suggested that an abnormal P-wave axis was associated with a pooled risk ratio of 2.12 (95% CI 1.49 to 3.01) for the detection of incident atrial fibrillation. This comprehensive systematic review and meta-analysis, indicates the positive association of abnormal P wave axis and future detection of AF. Utilisation of abnormal P-wave axis, alongside other parameters, may allow clinicians to better risk-stratify individuals at increased risk of AF, and thus identify those who may benefit most from prolonged cardiac monitoring or targeted anticoagulation.

Left Atrial Volumetric/Mechanical Coupling Index in the General Population: Distribution and Prediction of Incident Atrial Fibrillation: From the Copenhagen City Heart Study

Left Atrial Volumetric/Mechanical Coupling Index in the General Population: Distribution and Prediction of Incident Atrial Fibrillation: From the Copenhagen City Heart Study

Prediction of incident atrial fibrillation with cardiac biomarkers and left atrial volumes

ObjectiveAtrial fibrillation is a common arrhythmia associated with risk of stroke, heart failure and death. We aimed to elucidate the associations of cardiac biomarkers, echocardiographic left atrial volumetric indices and...

Prediction of incident atrial fibrillation with GDF-15 and echocardiographic left atrial volumes: data from the Akershus cardiac examination 1950 study

Abstract Background Left atrial structural remodeling is associated with the risk of incident atrial fibrillation, but there is uncertainty as to what echocardiographic left atrial indices that best predict incident atrial fibrillation. GDF-15 is a marker of oxidative stress and inflammation that is closely associated with the risk of stroke, bleeding and mortality in patients with atrial fibrillation. There is, however, limited data regarding its relation with left atrial remodeling and incident atrial fibrillation in the general population. Purpose To investigate the association between GDF-15 and left atrial volumes, and their prognostic properties for incident atrial fibrillation in a general population cohort. Methods The Akershus Cardiac Examination 1950 Study is a population-based cohort of individuals born in 1950. Comprehensive baseline examinations were performed for 3706 study participants in 2012–2015. Left atrial maximum and minimum volumes were measured, indexed to body surface area yielding LAVimax and LAVimin. Left atrial emptying fraction (LAEF) was calculated as ((LAVimax - LAVimin)/LAVimax)x100%. Individuals with known atrial fibrillation at baseline (n=166) were excluded, leaving 3540 included in this analysis. National health registries provided clinical outcome data regarding atrial fibrillation and death from any cause. Incident atrial fibrillation cases were validated by health records. Atrial flutter was included in the outcome definition. The CHARGE-AF risk score for atrial fibrillation was quantified. Prediction of incident atrial fibrillation was assessed by Cox proportional hazards regression models and visualised by Kaplan-Meier plots. Results Mean age at baseline was 63.9±0.7 years and 1763 (49.8%) were female. GDF-15 correlated weakly with LAVimax (r=−0.07, p&amp;lt;0.001) and LAVimin (r=−0.05 p=0.007), but not with LAEF (r=−0.01, p=0.61). During a median follow-up of 6.4 years, incident atrial fibrillation occurred in 135 (3.8%) individuals. Subjects who developed atrial fibrillation had larger LAVimax (30.1±8.3 vs. 26.2±6.9 mL/m2; p&amp;lt;0.001), larger LAVimin (17.1±7.1 vs. 14.1±4.7 mL/m2; p&amp;lt;0.001), and lower LAEF (44.3±10.7 vs. 46.3±9.0%; p=0.019). Concentrations of GDF-15 were not significantly different in those who experienced atrial fibrillation (median 850 [670–1107] vs. 790 [630–1020] ng/L; p=0.06). Left atrial volumes and LAEF, but not GDF-15, were associated with incident atrial fibrillation (Table 1 and Figure 1). Conclusions In this population-based cohort study, LAVimax, LAVimin, and LAEF predicted incident atrial fibrillation independently of the CHARGE-AF risk score. GDF-15 was not significantly associated with left atrial structural remodeling or with increased risk of incident atrial fibrillation. Funding Acknowledgement Type of funding sources: Public hospital(s). Main funding source(s): Vestre Viken HF, Hospital TrustAkershus University Hospital

Exploring circulating biomarkers for risk prediction of incident atrial fibrillation – insights from the BiomarCaRE project

Abstract Background Atrial fibrillation (AF) remains a major health issue in Europe and worldwide. Risk prediction is crucial to identify at-risk individuals and prevent subsequent complications of AF such as stroke and heart failure. Biomarker-enriched, personalized risk prediction offers great potential for population-wide prevention beyond traditional cardiovascular risk factors. Purpose We aimed to identify robust predictors for incident AF using classical regressions and machine learning (ML) techniques within a broad spectrum of candidate variables. Methods Three European community cohorts from the Biomarkers for Cardiovascular Risk Assessment in Europe (BiomarCaRE) consortium were included to explore the predictive utility of 14 biomarkers mirroring distinct pathophysiological pathways of AF including lipids, inflammation (C-reactive protein [CRP]), renal, and myocardium-specific markers (N-terminal pro B-type natriuretic peptide [NT-proBNP], high-sensitivity troponin I [hsTnI]) within a population-based sample of 42,280 individuals free of AF at baseline. Investigated biomarkers were examined in relation to incident AF using Cox regressions adjusted for multiple cardiovascular risk factors, and additionally by C-indices and net reclassification improvement (NRI) when compared to a reference model incorporating clinical variables. Their predictive utility for incident AF was further analyzed using different ML methods, including Least Absolute Shrinkage and Selection Operator (LASSO) and Random Survival Forest (RSF). Results Of 42,280 individuals (21,843 women [51.7%]; median [interquartile range, IQR] age, 52.2 [42.6, 62.0] years), 1496 (3.5%) developed AF during a median follow-up time of 5.7 years. In multivariable-adjusted Cox regression analysis, NT-proBNP was the strongest circulating predictor of incident AF (hazard ratio [HR] per standard deviation [SD] 1.93, 95% CI 1.82–2.04; P&amp;lt;0.001). Further, hsTnI (HR per SD 1.18, 95% CI 1.13–1.22; P&amp;lt;0.001), cystatin C (HR per SD 1.16, 95% CI 1.10–1.23; P&amp;lt;0.001) and CRP (HR per SD 1.08, 95% CI 1.02–1.14, P=0.012) correlated positively with new-onset AF. NT-proBNP enhanced model discrimination (ΔC-index 0.037, 95% CI 0.029–0.044) markedly and yielded the best reclassification improvement (NRI 0.237, 95% CI 0.187–0.287) when compared to the clinical model. Neither the addition of hsTnI to NT-proBNP, nor a model comprising all investigated biomarkers further increased discrimination or reclassification substantially. In different ML models, NT-proBNP and age were the strongest predictors of incident AF. Conclusions Using a dual approach with both classical regressions and modern ML methods, NT-proBNP consistently remained the strongest blood-based predictor of incident AF with relevant discriminative ability and reclassification yield beyond classical cardiovascular risk factors. The clinical benefit of these findings for AF risk prediction needs to be tested prospectively. Funding Acknowledgement Type of funding sources: Public grant(s) – EU funding. Main funding source(s): BiomarCaRE (FP7, HEALTH-F2-2011-278913)European Union's Horizon 2020 research and innovation programme (grant agreement number 847770, AFFECT-EU)

Morphology-Voltage-P-wave duration (MVP) score in prediction of incident atrial fibrillation in an elderly epidemiologic cohort

Abstract Background Interatrial block (IAB) on electrocardiography (ECG) and reduced P wave voltage reflect atrial abnormalities which may contribute to development of atrial fibrillation (AF). Purpose We aimed to assess the value of a recently proposed ECG risk score that combines the morphology, voltage and length of the P wave (MVP score) for prediction of incident AF in a prospective population-based setting. Material and methods The study population is based on the large, prospective Malmö Preventative Project (MPP) cohort. We included subjects without a history of AF, with a readable ECG in sinus rhythm and an echocardiography performed in 2002–2006 (n=983, mean age 70±5 years, 38% females). Median follow-up was 4.2 (IQR 3.7–4.8) years. ECGs were digitally processed using the Glasgow algorithm. Advanced IAB (aIAB) was defined as a P-wave ≥120 ms and biphasic morphology (+/−) in inferior leads, partial IAB (pIAB) as P-wave ≥120 ms and a monophasic positive morphoology in inferior leads. MVP score was calculated based on the P-wave morphology in inferior leads, voltage in lead 1, and P-wave duration (Table 1).Incident AF events (n=66, 7%) were obtained from the Swedish Hospital Discharge Register and Cause of Death Register. Cox regression analysis and Kaplan Meier curve analysis were used to study the association of echocardiographic and P-wave characteristics with the risk of new onset AF. Results At baseline the mean MVP score was 1±1, none of the subjects had MVP score above 4. MVP score correlated with left atrial (LA) area: Pearson r=0.192, p&amp;lt;0.001. After adjustment for age, gender and LA enlargement expressed as LA area &amp;gt;20 cm2, new onset AF was associated with MVP score 4 (HR 6.17, 95% CI 1.76–21.64 compared to those with MVP score &amp;lt;4, Figure 1). Neither aIAB (univariate HR 1.93 CI 95% 0.82–4.56), nor pIAB (univariate HR 1.16 CI 95% 0.52–2.55) predicted incident AF. Conclusion In a population-based elderly epidemiological cohort, the ECG-based MVP score was significantly associated with incident AF, independently of LA enlargement. Funding Acknowledgement Type of funding sources: Private grant(s) and/or Sponsorship. Main funding source(s): The Swedish Heart-Lung Foundation

Natural Language Processing to Improve Prediction of Incident Atrial Fibrillation Using Electronic Health Records.

BackgroundModels predicting atrial fibrillation (AF) risk, such as Cohorts for Heart and Aging Research in Genomic Epidemiology AF (CHARGE‐AF), have not performed as well in electronic health records. Natural language processing (NLP) may improve models by using narrative electronic health record text.Methods and ResultsFrom a primary care network, we included patients aged ≥65 years with visits between 2003 and 2013 in development (n=32 960) and internal validation cohorts (n=13 992). An external validation cohort from a separate network from 2015 to 2020 included 39 051 patients. Model features were defined using electronic health record codified data and narrative data with NLP. We developed 2 models to predict 5‐year AF incidence using (1) codified+NLP data and (2) codified data only and evaluated model performance. The analysis included 2839 incident AF cases in the development cohort and 1057 and 2226 cases in internal and external validation cohorts, respectively. The C‐statistic was greater (P<0.001) in codified+NLP model (0.744 [95% CI, 0.735–0.753]) compared with codified‐only (0.730 [95% CI, 0.720–0.739]) in the development cohort. In internal validation, the C‐statistic of codified+NLP was modestly higher (0.735 [95% CI, 0.720–0.749]) compared with codified‐only (0.729 [95% CI, 0.715–0.744]; P=0.06) and CHARGE‐AF (0.717 [95% CI, 0.703–0.731]; P=0.002). Codified+NLP and codified‐only were well calibrated, whereas CHARGE‐AF underestimated AF risk. In external validation, the C‐statistic of codified+NLP (0.750 [95% CI, 0.740–0.760]) remained higher (P<0.001) than codified‐only (0.738 [95% CI, 0.727–0.748]) and CHARGE‐AF (0.735 [95% CI, 0.725–0.746]).ConclusionsEstimation of 5‐year risk of AF can be modestly improved using NLP to incorporate narrative electronic health record data.

A new angle on atrial fibrillation risk prediction; the role of P-wave axis in future AF

Abstract Funding Acknowledgements Type of funding sources: None. Background/Introduction There is growing interest in the prediction of incident atrial fibrillation. This is driven by the vast amounts of data in electronic records that may act as a substrate for prediction, and has led to demographic, biochemical and electrophysiological strategies (figure 1a). The 12-lead ECG has been a particularly rich target for possible prediction strategies, with a variety of individual parameters and even artificial intelligence methods being examined. Purpose The P-wave axis is an ECG parameter that reflects the dominant vector of atrial depolarisation. The normal P-wave axis is considered to be 0-75º. There is a large body of literature suggesting that AF reflects structural and conduction abnormalities of the atria, and thus the P-wave axis may represent a sensitive parameter to detect such changes, prior to the onset of AF (figure 1b). Methods A systematic review and meta-analysis of published literature associating abnormal P-wave axis and the development of incident AF was performed. The electronic databases PUBMED, EMBASE, Web of Science and Cochrane were systematically searched using the key words "P-wave axis" and "P wave axis" from inception to October 2021. An initial titular analysis, followed by abstract and full text screening was performed. Data was extracted using a pre-specified collection form. A random-effects model with generic inverse variance weights was utilised to pool the most adjusted effect measure from each paper. A funnel plot was used to assess publication bias. Results After excluding duplicate studies, 568 studies were screened. A total of twelve studies were identified that associated an abnormal p-wave axis with the subsequent detection of atrial fibrillation. The eight studies that considered abnormal P-wave axis as being &amp;lt;0º or &amp;gt;75º were pooled for meta-analysis. In the pooled studies a total of 78,222 patients were included with 5656 cases of incident atrial fibrillation identified. The meta-analysis of the studies suggested that an abnormal P-wave axis was associated with a pooled risk ratio of 2.10 (95% CI 1.48 to 2.72) for the detection of incident atrial fibrillation (figure 2). Sensitivity analysis by removing one study at each time did not alter the results. Conclusion This comprehensive systematic review and meta-analysis including 78,327 patients from 8 studies, indicates the positive association of abnormal P wave axis and future detection of AF. It thus confirms a two-fold risk of abnormal p wave axis in parallel with other established factors such as hypertension, age and abnormal systolic heart function. Utilisation of abnormal p wave axis, can allow clinicians to better risk-stratify individuals at increased risk of AF. This may lead to more tailored and precision strategies for prolonged cardiac monitoring or targeted anticoagulation approaches.