Automatic detection of distorted plethysmogram pulses in neonates and paediatric patients using an adaptive-network-based fuzzy inference system

摘要: Despite the fact that pulse oximetry has become an essential technology in respiratory monitoring of neonates and paediatric patients, it is still fraught with artefacts causing false alarms resulting from patient or probe movement. As shape plethysmogram always been considered as a useful visual indicator for determining reliability SaO"2 numerical readings, automation this observation might benefit health care providers at bedside. We observed systolic upstroke time (t"1), diastolic (t"2) heart rate (HR) extracted constitute features, which can be used detecting normal distorted pulses. developed technique classifying pulses into two categories: valid artefact via implementations fuzzy inference systems (FIS), were tuned using adaptive-network-based system (ANFIS) receiver operating characteristics (ROC) curves analysis. Features total 22,497 waveforms obtained 13 patients to systematically optimise FIS. A further 2843 another eight testing system, visually classified 1635 (58%) 1208 (42%) segments. For optimum area under ROC curve was 0.92. The able classify 1418 (87%) segments 897 (74%) correctly. calculations system's performance showed 87% sensitivity, 81% accuracy 74% specificity. In comparison 95% confidence interval (CI) thresholding method, higher specificity (P=0.008,P 0.05) (P=0.053,P>0.05). therefore conclude algorithm some potential pulse. However, evaluation needed larger groups.