A big data approach to myocyte membrane analysis: Using populations of models to understand the cellular causes of heart failure

Experimentally-calibrated populations of models (ePoMs) allow to elucidate the trends hidden behind large amounts of data, but they have never been used as a means to monitor the temporal evolution of a heart's electrical activity. This work aimed at using ePoMs to understand the cell membrane anomalies that lead to heart failure. A population of the Tusscher-Noble-Noble-Panfilov model was calibrated to the activation-recovery intervals measured from epicardial electrograms acquired during a Physio-heart experiment. A Mann-Whitney-Wilcoxon U-test was performed on the statistical distributions of the calibrated parameters, at different time points during the experiment, to elucidate the physiology changes that would have led to the resulting ePoMs. The methodology developed in this paper could detect the specific pathological ion dynamics responsible for the abnormal electrical behavior observed during the experiment. Furthermore, the analysis of the electrical activities was capable of detection of pathologies at an earlier stage when compared to the analysis of the cardiac output alone. The use of big data analytics proved to be more effective than the traditional signal analysis approach in predicting heart failure; additionally, this approach accounts for variabilities in both the healthy and the pathological conditions.