“…For instance, if we have computed the thresholds utilizing only tapes 100, 101, 103, 113, 115, 122, 123, our circuit performance would be over 99.9 (%) for selectiveness (Se) and positive predictivity (P + ), and less than 0.1 (%) for failed detection (F d ). Also, other works, such as [12] only use real data directly acquired by their own systems which makes unfeasible a fair comparison. On the other hand, we do have some works, such as in [13], which present better results than ours, even for the whole set of MIT-BIH tapes.…”