Algorithm fusion for the early detection of apnea-bradycardia in preterm infants

Episodes of apnea-bradycardia are frequent in preterm infants. The incidence and severity of these events may lead to neurological morbidity or even to the infant death. Even if algorithms for bradycardia detection have been developed, they are inefficient and usually produce false or late alarms. In this work, a new algorithm for the detection of apnea-bradycardia in preterm infants is proposed, based on the fusion of different detection algorithms. A quantitative evaluation of the proposed algorithm and a comparison with two other algorithms proposed in the literature is presented. A database of 40 newborns, with a total of 1188 episodes of apnea-bradycardia was used for the evaluation. The proposed algorithm presents sensitivity and specificity of 97.67% and 97.00%, respectively. Furthermore, the mean time- delay for event detection is decreased to 2.22 beats, which is lower than the mean of the algorithms proposed in literature.