Aims: Atrial fibrillation (AF) and heart failure often co-exist. Early identification of AF-patients at risk for AF-induced cardiomyopathies (AF-CMP) is desirable to reduce both morbidity and mortality as well as health care costs. We aimed to investigate the characteristics of beat-to-beat-patterns in AF to prospectively discriminate AF-patients with and without AF-CMP.

Methods: A dataset of 10,234 5-minute length RR-intervals derived from 26 AF-CMP-patients and 26 control patients and extracted from 1-lead Holter-ECGs was used in the analysis. A total of 14 features was extracted and a decision tree classifier with 5-fold cross-validation technique was trained, validated, and tested on the dataset randomly split after selecting the most informative features. The derived algorithm was then tested on 2,261 5-min segments from six AF-CMP- and six control-patients and validated for various time points.

Results: The algorithm based on the spectral entropy of the RR-interval, the mean value of the relative RR-interval, and the root mean square of successive differences of the relative RR-interval yielded an accuracy of 73.51%, specificity of 91.38%, sensitivity of 64.67%, and PPV of 87.00% to correctly stratify segments to AF-CMP or control (Figure 1A, correctly classified patients in green). Considering the majority vote of the segments of each patient, 10 out of 12 patients (83,33%) were correctly classified (Figure 1B).

Conclusion: Beat-to-beat-analysis using a machine learning classifier identifies patients with AF-induced cardiomyopathy with clinically relevant diagnostic properties. Application of this algorithm to routine care may improve early identification of patients at risk for AF-induced cardiomyopathy.