Kraaijenga test was implemented on 234 apnea episodes,the

Kraaijenga et al.’s study showed the relationship betweenCI and dEMG7, which has proved the feasibility andrepeatability of dEMG to monitor respiratory rate (RR) andheart rate (HR) in preterm infants.Cohen et al.’s work has shown the multi-model approachcan be used to auto-detect sleep apnea in infants10, wherefeatures extracted from electrocardiography (ECG) and SpO2are used to train a classifier. Although the performanceof their system failed to reach the requirement of clinicaluse, their work showed it is possible to auto detect theapnea based on low-cost and minimally invasive diagnosticsystems.Lee et al. has developed an algorithm to detect centralapnea using ECG and CI5, the cardiac artifact was filteredout from the chest impedance and RR interval was used asclock for re-sampled CI, then the running standard deviationwas calculated to represent the probability of apnea, whichcan be used as the standard to judge the occurrence ofapnea. The test was implemented on 234 apnea episodes,the accuracy was as high as 91%.Kraaijenga et al.’s study stated that dEMG is able toclassify preterm apnea and improve the accuracy of apneaclassification compared to CI11. Their study was based onvisually classification, which means classifying apnea by 22assessors. It did cause much labor. In addition, they onlyselected apnea from stable data, which did not take unstabledata into consideration. Their work also showed a criterionto classify apena, using only stable tracing of flow, SpO2 andHR, a cessation of breathing in the flow signal for more than20s or of shorter duration if accompanied by hypoxemia orbradycardia11.Zhang et al. used ensemble empirical mode decomposition(EMD), a kind of non-stationary signal analyze methodintroduced by Huang et al.12, to eliminate the noise insurface EMG. Power line interference, white Gaussian noise(WGN), and motion artifact or baseline wandering (BW)were discussed13. Their algorithm worked well but wastime consuming and may not suitable to neonatal signal.Gating method is one of the most popular methods toeliminate ECG interference that raised by O’Brien et al.14.Several filters are used to eliminate P and T waves, local peakdetector was designed to find the QRS complex. It is reportedin Bartolo et al.’s study that gating method was feasible toeliminate ECG interference from diaphragm EMG15.