Resumen

SAEZ, Patricio  y  HERRERA, Carlos. Proposal of a Markov chains based method for predicting paroxysmal atrial fibrillations. Ingeniare. Rev. chil. ing. [online]. 2015, vol.23, n.3, pp.449-457. ISSN 0718-3305.  http://dx.doi.org/10.4067/S0718-33052015000300014.

Atrial fibrillation has been one of the most studied arrhythmias due to its high rate of recurrence. Given the importance of this condition it is possible to find in the literature several works based on the detection of ectopic beats and RR intervals among others. This paper addresses this issue supported by the detection of atrial fibrillation, centered on the use of the proposed method, based on Markov Chains. The method identifies states from two series corresponding to the RR and PR intervals of time. The results were validated using a control set from the international competition of paroxysmal atrial fibrillation prediction. The method had a sensitivity of 82.14 % (23/28 episodes in the study) and a predictive factor of 79.31%. Validation of the results was complemented with other cases from the database MIT-BIH atrial fibrillation with a sensitivity of 81.15% (112/138 episodes in the study). The experimental evidence shows that the proposed method outperforms the prediction obtained by other methods found in the literature.

Palabras clave : Proxysmal atrial fibrillation; Markov chains; electrocardiogram; time series; ergodicity.

        · resumen en Español     · texto en Español     · Español ( pdf )