The method of using deep learning technology to realize automatic sleep staging needs a lot of data support, and its computational complexity is also high. In this paper, an automatic sleep staging method based on power spectral density (PSD) and random forest is proposed. Firstly, the PSDs of six characteristic waves (K complex wave, δ wave, θ wave, α wave, spindle wave, β wave) in electroencephalogram (EEG) signals were extracted as the classification features, and then five sleep states (W, N1, N2, N3, REM) were automatically classified by random forest classifier. The whole night sleep EEG data of healthy subjects in the Sleep-EDF database were used as experimental data. The effects of using different EEG signals (Fpz-Cz single channel, Pz-Oz single channel, Fpz-Cz + Pz-Oz dual channel), different classifiers (random forest, adaptive boost, gradient boost, Gaussian naïve Bayes, decision tree, K-nearest neighbor), and different training and test set divisions (2-fold cross-validation, 5-fold cross-validation, 10-fold cross-validation, single subject) on the classification effect were compared. The experimental results showed that the effect was the best when the input was Pz-Oz single-channel EEG signal and the random forest classifier was used, no matter how the training set and test set were transformed, the classification accuracy was above 90.79%. The overall classification accuracy, macro average F1 value, and Kappa coefficient could reach 91.94%, 73.2% and 0.845 respectively at the highest, which proved that this method was effective and not susceptible to data volume, and had good stability. Compared with the existing research, our method is more accurate and simpler, and is suitable for automation.
Sleep apnea (SA) detection method based on traditional machine learning needs a lot of efforts in feature engineering and classifier design. We constructed a one-dimensional convolutional neural network (CNN) model, which consists in four convolution layers, four pooling layers, two full connection layers and one classification layer. The automatic feature extraction and classification were realized by the structure of the proposed CNN model. The model was verified by the whole night single-channel sleep electrocardiogram (ECG) signals of 70 subjects from the Apnea-ECG dataset. Our results showed that the accuracy of per-segment SA detection was ranged from 80.1% to 88.0%, using the input signals of single-channel ECG signal, RR interval (RRI) sequence, R peak sequence and RRI sequence + R peak sequence respectively. These results indicated that the proposed CNN model was effective and can automatically extract and classify features from the original single-channel ECG signal or its derived signal RRI and R peak sequence. When the input signals were RRI sequence + R peak sequence, the CNN model achieved the best performance. The accuracy, sensitivity and specificity of per-segment SA detection were 88.0%, 85.1% and 89.9%, respectively. And the accuracy of per-recording SA diagnosis was 100%. These findings indicated that the proposed method can effectively improve the accuracy and robustness of SA detection and outperform the methods reported in recent years. The proposed CNN model can be applied to portable screening diagnosis equipment for SA with remote server.
【摘要】 目的 探讨成都市成华区中老年人群血脂水平、分布特点及其与胰岛素抵抗指数（HOMA-IR）的关系。 方法 2007年5月在此区中老年（50~79岁）人群中随机抽取672人进行心血管危险因素研究调查，对其血脂水平及HOMA-IR进行统计分析。 结果 人群当中①女性各血脂项目的水平均比男性高，其中总胆固醇（TC）、高密度脂蛋白胆固醇（HDL-C）的差异有统计学意义（Plt;0.05）；②三酰甘油（TG）升高的比例较高，其中男性为30.0%，女性为27.6%；大部分人群HDL-C、低密度脂蛋白胆固醇（LDL-C）水平处于合适范围，HDL-C降低的比例为6.0%，LDL-C升高的比例为7.3%；③随着TG水平的升高、HDL-C 水平的降低，HOMA-IR呈升高趋势；LDL-C水平的升高，HOMA-IR呈升高趋势，仅在女性人群中差异有统计学意义（Plt;0.05），在男性人群中差异无统计学意义；④TG与HOMA-IR呈正相关，相关系数为0.185（P=0.000）；HDL-C与HOMA-IR呈负相关，相关系数为-0.145（P=0.000）。LDL-C与HOMA-IR呈正相关，相关系数为0.099（P=0.010）。 结论 TG增高是成都市成华区中老年人群的显著特点，女性HDL-C比男性高；血脂紊乱与胰岛素抵抗相关。【Abstract】 Objective To investigate the relationship between blood lipids level and homeostasis model assessment-insulin resistance (HOMA-IR) in elder people in Chengdu. Methods In May 2007, 672 people aged from 50 to 79 years in Chengdu were recruited by random sampling methods for the survey of cardiovascular risk factors. The blood lipids level and HOMA-IR were statistically analyzed. Results ① The serum total cholesterol (TC) and high density lipoprotein chole sterol (HDL-C) were obviously higher in women than those in men (Plt;0.05). ② Triacylglycerol (TG) increased in 30.0% of men and 27.6% of women; HDL-C and low density lipoprotein cholesterin (LDL-C) in most of the involved people were appropriate. ③ HOMA-IR increased as the TG level increased and HDL-C decreased; HOMA-IR increased as the LDL-C level increased, which was significant in the females (Plt;0.05). ④ HDL-C was positively correlated with HOMA-IR (r=-0.145, P=0.000); LDL-C was positively correlated with HOMA-IR (r=0.099, P=0.010). Conclusion The increase of hypertriglyceridemia was the most frequent type of the dislipidemia in the elder people in Chengdu; HDL-C level is higher in women than in men. Dyslipidemia is correlated with insulin resistance.
【摘要】 目的 分析成都地区中老年居民脉压（pulse pressure， PP）及脉压指数（pulse presure index，PPI）与高尿酸血症（hyperuricemia，HUA）的关系。 方法 利用2007年5月代谢综合征研究调查资料（共1 061人），依据PP[≤60 mm Hg（1 mm Hg=0.133 kPa）、gt;60 mm Hg]和PPI（≤0.450、gt;0.450）将人群分为正常组及增高组，分析两组人群尿酸水平及HUA患病率，采用单因素回归及logistic回归分析PP及PPI与HUA关系。 结果 ①PP/PPI增高组血浆尿酸水平明显高于PP/PPI正常组，差异有统计学意义（P=0.000）。②PP/PPI增高组HUA患病率明显高于PP/PPI正常组，差异有统计学意义（P=0.026、0.027）。③单因素回归和logistic回归分析皆提示PP及PPI与HUA呈正相关。 结论 成都地区PP及PPI与血浆尿酸水平关系密切，PP/PPI增高可能是HUA的危险因素。【Abstract】 Objective To evaluate the relationship between pulse pressure (PP), pulse pressure index (PPI) and hyperuricemia (HUA) among middle-aged and aged residents in Chengdu. Methods Based on the level of PP [≤60 mm Hg (1 mm Hg=0.133 kPa),gt;60 mm Hg] and PPI (≤0.450,gt;0.450), We divided the 1 061 middle-aged or aged people into normal PP/PPI group and augmented PP/PPI group. All patients came from the survey for metabolic syndrome study in May 2007. We analyzed the distribution of serum uric acid (UA) and HUA, and analyzed the relationship between PP, PPI and HUA by using single-factor and logistic regression analysis. Results The index of UA in the augmented PP/PPI groups was higher than that in the normal groups with a significant difference (P=0.000). The prevalence of HUA in the augmented PP/PPI groups was statistically higher than that in the normal groups (P=0.026, 0.027). Single-factor and logistic regression analysis showed that PP and PPI were both positively correlated to HUA. Conclusion The abnormalities of PP and PPI are closely related to metabolism disorder in Chengdu, and high level of PP or PPI is probably risk factors for HUA.
Cognitive impairment is one of the three primary symptoms of schizophrenic patients and shows important value in early detection and warning for high-risk individuals. To study the specifics of electroencephalogram (EEG) in patients with schizophrenia under the cognitive load, we collected EEG signals from 17 schizophrenic patients and 19 healthy controls, extracted signals of each band based on wavelet transform, calculated the characteristics of nonlinear dynamic and functional brain networks, and automatically classified the two groups of people by using a machine learning algorithm. Experimental results indicated that the correlation dimension and sample entropy showed significant differences in α, β, θ, and γ rhythm of the Fp1 and Fp2 electrodes between groups under the cognitive load. These results implied that the functional disruptions in the frontal lobe might be the important factors of cognitive impairments in schizophrenic patients. Further results of the automatic classification analysis indicated that the combination of nonlinear dynamics and functional brain network properties as the input characteristics of the classifier showed the best performance, with the accuracy of 76.77%, sensitivity of 72.09%, and specificity of 80.36%. The results of this study demonstrated that the combination of nonlinear dynamics and function brain network properties may be potential biomarkers for early screening and auxiliary diagnosis of schizophrenia.
A great number of studies have demonstrated the structural and functional abnormalities in chronic schizophrenia (SZ) patients. However, few studies analyzed the differences between first-episode, drug-naive SZ (FESZ) patients and normal controls (NCs). In this study, we recruited 44 FESZ patients and 56 NCs, and acquired their multi-modal magnetic resonance imaging (MRI) data, including structural and resting-state functional MRI data. We calculated gray matter volume (GMV), regional homogeneity (ReHo), amplitude of low frequency fluctuation (ALFF), and degree centrality (DC) of 90 brain regions, basing on an automated anatomical labeling (AAL) atlas. We then applied these features into support vector machine (SVM) combined with recursive feature elimination (RFE) to discriminate FESZ patients from NCs. Our results showed that the classifier using the combination of ReHo and ALFF as input features achieved the best performance (an accuracy of 96.97%). Moreover, the most discriminative features for classification were predominantly located in the frontal lobe. Our findings may provide potential information for understanding the neuropathological mechanism of SZ and facilitate the development of biomarkers for computer-aided diagnosis of SZ patients.