Exploring mortality representation and the impact of COVID-19 in Modena: insights from “An ECG-based machine-learning approach for mortality risk assessment in a large European population”

We read with interest and appreciate Mc Loughlin's reply [1] to our recent work “An ECG-based machine-learning approach for mortality risk assessment in a large European population” [2], and we would like to address the raised points. As stated in the “Limitations” Section of [2], the rate of deceased patients in the considered sample is over-representative of the overall population mortality, and the two should not be compared. This is because only patients who survived for at least 5 years or who died within 5 years from ECG registration were included. In other words, surviving patients with an observation period shorter than 5 years and deceased after 5 years were excluded. In addition, only adult patients from 40 to 90 years of age were enrolled and precise ECG exclusion criteria were applied: therefore, the measured adverse event rate cannot be compared to the mortality of the resident population. Finally, we would like to point out that the rate of deceased was similar to the ones found in previous studies with the same scope and a long follow-up period, as referenced in [2].