Heart Alert ECG Sonification


One of the challenges when analyzing ECG recordings is to determine which parts of the signal are meaningful data, and which others are the result of noises and artifacts. Moreover, a difficulty for accurate diagnostics is given by the fact that an abnormality in the ECG might be present in only a specific subset of leads, while other leads remain closer to a healthy signal. Sonification appears to be a good approach to help clinicians in the analysis and identification of important variations of the signal. First, because the human ear has the capability to rapidly and robustly detect changes and patterns even in noisy signals, and second, because by giving the users the possibility to interact with the resulting sonifications, they could refine the sonifications themselves in order to enhance any pattern they regard as relevant. Heart Alert is an interactive setting where the users can upload a dataset and interactively adjust sonification parameters, for instance in search of settings that optimize the contrast between a baseline (regular) and abnormal (pathologic) case, based on pre-recorded real ECG data sets.

Involved People




Supplementary Material