Methods and systems are provided for locating arrhythmogenic cardiac focus. A training set of electrical signal data is acquired and automatically analyzed to identify CFAE locations. The data representing the CFAE locations are and subjected to determinations of medical significance, such as ablation decisions, by a battery of physicians. The decisions and descriptive statistics of the data are fitted to a binary logistic mixed model, in which the physician decisions are treated as random effects. The model is then applied to patient CFAE data to establish estimates of medical significance, such as suitability for ablation.