Prediction of early death after atrial fibrillation diagnosis using a machine learning approach: A French nationwide cohort study

AIMS

Atrial fibrillation is associated with important mortality but the usual clinical risk factor based scores only modestly predict mortality. This study aimed to develop machine learning models for the prediction of death occurrence within the year following atrial fibrillation diagnosis and compare predictive ability against usual clinical risk scores.

METHODS AND RESULTS

We used a nationwide cohort of 2,435,541 newly diagnosed atrial fibrillation patients seen in French hospitals from 2011 to 2019. Three machine learning models were trained to predict mortality within the first year using a training set (70% of the cohort). The best model was selected to be evaluated and compared with previously published scores on the validation set (30% of the cohort). Discrimination of the best model was evaluated using the C index. Within the first year following atrial fibrillation diagnosis, 342,005 patients (14.4%) died after a period of 83 (SD 98) days (median 37 [10-129]). The best machine learning model selected was a deep neural network with a C index of 0.785 (95% CI, 0.781-0.789) on the validation set. Compared to clinical risk scores, the selected model was superior to the CHA2DS2-VASc and HAS-BLED risk scores and superior to dedicated scores such as Charlson Comorbidity Index and Hospital Frailty Risk Score to predict death within the year following atrial fibrillation diagnosis (C indexes: 0.597; 0.562; 0.643; 0.626 respectively. P < .0001).

CONCLUSION

Machine learning algorithms predict early death after atrial fibrillation diagnosis and may help clinicians to better risk stratify atrial fibrillation patients at high risk of mortality.

Phenotypes and outcomes in non-anticoagulated patients with atrial fibrillation: An unsupervised cluster analysis

BACKGROUND

Patients with atrial fibrillation are characterized by great clinical heterogeneity and complexity. The usual classifications may not adequately characterize this population. Data-driven cluster analysis reveals different possible patient classifications.

AIMS

To identify different clusters of patients with atrial fibrillation who share similar clinical phenotypes, and to evaluate the association between identified clusters and clinical outcomes, using cluster analysis.

METHODS

An agglomerative hierarchical cluster analysis was performed in non-anticoagulated patients from the Loire Valley Atrial Fibrillation cohort. Associations between clusters and a composite outcome comprising stroke/systemic embolism/death and all-cause death, stroke and major bleeding were evaluated using Cox regression analyses.

RESULTS

The study included 3434 non-anticoagulated patients with atrial fibrillation (mean age 70.3±17 years; 42.8% female). Three clusters were identified: cluster 1 was composed of younger patients, with a low prevalence of co-morbidities; cluster 2 included old patients with permanent atrial fibrillation, cardiac pathologies and a high burden of cardiovascular co-morbidities; cluster 3 identified old female patients with a high burden of cardiovascular co-morbidities. Compared with cluster 1, clusters 2 and 3 were independently associated with an increased risk of the composite outcome (hazard ratio 2.85, 95% confidence interval 1.32-6.16 and hazard ratio 1.52, 95% confidence interval 1.09-2.11, respectively) and all-cause death (hazard ratio 3.54, 95% confidence interval 1.49-8.43 and hazard ratio 1.88, 95% confidence interval 1.26-2.79, respectively). Cluster 3 was independently associated with an increased risk of major bleeding (hazard ratio 1.72, 95% confidence interval 1.06-2.78).

CONCLUSION

Cluster analysis identified three statistically driven groups of patients with atrial fibrillation, with distinct phenotype characteristics and associated with different risks for major clinical adverse events.

Clinical Phenotypes and Atrial Fibrillation Recurrences after Catheter Ablation: An Unsupervised Cluster Analysis

Catheter ablation (CA) is a well-established treatment of atrial fibrillation (AF). Data-driven cluster analysis is able to better distinguish prognostically-relevant phenotype clusters among patients with AF. We performed a hierarchical cluster analysis in a cohort of AF patients undergoing a first CA and evaluate associations between identified clusters and recurrences of arrhythmia following ablation. The study included 209 AF patients treated with CA. 3 clusters with distinct characteristics were identified. Recurrences at one year occurred in 27.2% in Cluster 1, 43.2% in Cluster 2 and 60.9% in Cluster 3 (p<0.0001). Cluster classification was independently associated with arrhythmia recurrences (HR 1.58, 95% CI 1.01-2.49, p=0.046) after adjustment for age, CHA₂DS₂-VASc score, left atrial volume, type of atrial fibrillation and ejection fraction. To concluded, cluster analysis identified three statistically-driven groups among AF patients treated with CA with different risks for arrhythmia recurrences.

Mathematical modelling of haemorrhagic transformation within a multi-scale microvasculature network

Objective Haemorrhagic transformation (HT) is one of the most common complications after ischaemic stroke caused by damage to the blood-brain barrier (BBB) that could be the result of stroke progression or a complication of stroke treatment with reperfusion therapy. The aim of this study is to develop further a previous simple HT mathematical model into an enlarged multi-scale microvasculature model in order to investigate the effects of HT on the surrounding tissue and vasculature. In addition, this study investigates the relationship between tissue displacement and vascular geometry. Approach By modelling tissue displacement, capillary compression, hydraulic conductivity in tissue and vascular permeability, we establish a mathematical model to describe the change of intracranial pressure (ICP) surrounding the damaged vascular bed after HT onset applied to a 3D multi-scale microvasculature. The use of a voxel-scale model then enables us to compare our HT simulation with available clinical imaging data for perfusion and cerebral blood volume (CBV) in the multi-scale microvasculature network. Main results We showed that the haematoma diameter and the maximum tissue displacement are approximately proportional to the diameter of the breakdown vessel. Based on the voxel-scale model, we found that perfusion reduces by approximately 13-17 % and CBV reduces by around 20-25 % after HT onset due to the effect of capillary compression caused by increased interstitial pressure. The results are in good agreement with the limited experimental data. Significance This model, by enabling us to bridge the gap between the microvascular scale and clinically measurable parameters, thus provides a foundation for more detailed validation and understanding of HT in patients.

Screening cultures for detection of methicillin-resistant Staphylococcus aureus in a population at high risk for MRSA colonisation: identification of optimal combinations of anatomical sites

This retrospective study analysed the diagnostic yield of single-site, two-site, and three-site anatomical surveillance cultures in a population of 4,769 patients at high risk for methicillin-resistant Staphylococcus aureus (MRSA) colonisation. Cultures of seven anatomical sites were used as the gold standard against which to measure the sensitivity of MRSA detection. Detection rates for the seven single-sites, 21 two-site, and 35 three-site combinations are presented. Single-site swabbing only detected 50.5% (nose) of total cases, while three-site surveillance achieved a 92% (groin + nose + throat) sensitivity of detection at best. It is recommended that at least three anatomical sites should be screened for MRSA colonisation in these high-risk patients.