exploration of reliable parameters scored by automated analysis in polysomnography

Background and objective: polysomnography (psg) is the gold standard for diagnosis of sleep disorders. several software programs are available to analyze sleep tests according to available guidelines and decrease the time and cost of psg analysis. this study aimed to compare the parameters of automated analyzer software with analysis of trained technician (manual analysis). materials and methods: twenty patients who underwent full-night psg were randomly selected. a sleep technologist who was blind to the study, scored sleep stages and respiratory events according to recommended criteria of american academy of sleep medicine (aasm) 2013, then an auto analysis was done using n-7000 amplifier. results of auto analysis and manual analysis were compared. descriptive statistics and paired t-test were used for data analysis. results: total sleep time (tst) and sleep efficiency (se) calculated by auto analysis was significantly more than manual analysis (511.82 ± 35.34 vs. 396.85 ± 75.97 for tst and 95.47 ± 3.74 vs. 74.14 ± 35.34 for se, respectively). furthermore, there was no concordance for sum of apneas and hypopneas during tst. however, calculated number of hypopneas in non-rapid eye movement (nrem) stage in auto analysis and manual analysis was quite similar. the least precision was observed in scoring of stages 3 and rem for auto analysis scoring and the most similarity for scoring of stage n2. conclusion: detecting hypopneas in nrem stage by auto analysis maybe the reliable parameter that could help the technicians during analysis of sleep test. there is a need for more advanced automated algorithms. furthermore, manual analysis is superior to automated one in psg analysis according to the current results.