OSAHS Diagnosis Based On Classify Of Blood Oxygen Saturation Signal

Due to high prevalence,high risk and high death rate,obstructive sleep apnea-hypopnea syndrome(OSAHS)has drawn more and more attention.To find an efficient and valid alternative of polysomnography(PSG),this paper uses the blood oxygen saturation(Sp O2)signal to count the number of apnea or hypopnea events during sleep,and then calculates the apnea hypopnea index(AHI)and evaluates OSAHS severity classification.In this paper,the Sp O2 signal collected by the finger-clip oximeter,both signals are segmented into 1-min episodes in the preprocessing step.This paper extracts 35 features based on time domain for the segmented signal,such as approximate entropy,central tendency measure(CTM)and LZ complexity,which are used to characterize the complexity of the signal,the degree of decline and duration of ? index,oxygen saturation decline index(ODI)and TSA characteristics,this paper and based on the principle of oxygen saturation index and extraction methods to summarize the ODIxy feature parameters.In this paper,feature selection methodology includes the Pearson correlation coefficient selection,the minimal redundancy maximal relevance(m RMR)algorithm selection and the wrapper based on backward selection algorithm,reducing the feature set from 35 to 7 dimensions.By comparing the training and testing of the classifier,according to the classification accuracy considerations,this paper selects the random forest classifier,which achieves the recognition accuracy of whether the "apnea-hypopnea" event is included in a segment of Sp O2 signal is 86.92%,the specificity is 90.7%.In this paper,a random forest classifier was used to calculate the AHI index of 25 subjects,and a linear regression analysis was performed with the AHI index obtained from PSG.This paper can achieve 92% accuracy in evaluates OSAHS severity classification.