The study is not worth paying attention to.

The datasets for positive cases and negative cases come from different databases. n=30 patients, on top of it.

All this does is recognize the patient/ECG technician who recorded the data. It's basically certain it doesnt generalize