For speech detection in Parkinson’s patients, we proposed a method based on time-frequency domain gradient statistics to analyze speech disorders of Parkinson’s patients. In this method, speech signal was first converted to time-frequency domain (time-frequency representation). In the process, the speech signal was divided into frames. Through calculation, each frame was Fourier transformed to obtain the energy spectrum, which was mapped to the image space for visualization. Secondly, deviations values of each energy data on time axis and frequency axis was counted. According to deviations values, the gradient statistical features were used to show the abrupt changes of energy value in different time-domains and frequency-domains. Finally, KNN classifier was applied to classify the extracted gradient statistical features. In this paper, experiments on different speech datasets of Parkinson’s patients showed that the gradient statistical features extracted in this paper had stronger clustering in classification. Compared with the classification results based on traditional features and deep learning features, the gradient statistical features extracted in this paper were better in classification accuracy, specificity and sensitivity. The experimental results show that the gradient statistical features proposed in this paper are feasible in speech classification diagnosis of Parkinson’s patients.
Multivariate time series problems widely exist in production and life in the society. Anomaly detection has provided people with a lot of valuable information in financial, hydrological, meteorological fields, and the research areas of earthquake, video surveillance, medicine and others. In order to quickly and efficiently find exceptions in time sequence so that it can be presented in front of people in an intuitive way, we in this study combined the Riemannian manifold with statistical process control charts, based on sliding window, with a description of the covariance matrix as the time sequence, to achieve the multivariate time series of anomaly detection and its visualization. We made MA analog data flow and abnormal electrocardiogram data from MIT-BIH as experimental objects, and verified the anomaly detection method. The results showed that the method was reasonable and effective.
There are so many biomechanical risk factors related with glaucoma and their relationship is much complex. This paper reviewed the state-of-the-art research works on glaucoma related mechanical effects. With regards to the development perspectives of studies on glaucoma biomechanics, a completely novel biomechanical evaluation factor -- Fractional Flow Reserve (FPR) for glaucoma was proposed, and developing clinical application oriented glaucoma risk assessment algorithm and application system by using the new techniques such as artificial intelligence and machine learning were suggested.
ObjectiveTo establish an appropriate diabetic retinopathy (DR) risk assessment model for patients with type 2 diabetes mellitus (T2DM).MethodsA retrospective clinical analysis. From January 2016 to December 2017, 753 T2DM patients in the Third Affiliated Hospital of Southern Medical University were analyzed retrospectively. Digital fundus photography was taken in all patients. Fasting plasma glucose (FPG), HbA1c, total bilirubin (TB), blood platelet, total cholesterol (TC), triglyceride (TG), high density lipoprotein cholesterol (HDL-c), low density lipoprotein cholesterol (LDL-c), apolipoprotein-A (apoA), apolipoprotein-B (apoB), serum creatinine, blood urea nitrogen (BUN), blood uric acid, fibrinogen (Fg), estimated glomerular filtration (eGFR) were collected. The patients were randomly assigned to model group and testify group, each had 702 patients and 51 patients respectively. Logistic regression was used to screen risk factors of DR and develop an assessment scale that can be used to predict DR. Goodness of fit was examined using the Hosmer-Lemeshow test and the area under the receiver operating characteristic (ROC) curve.ResultsAmong 702 patients in the model group, 483 patients were DR, 219 patients were NDR. The scores for DR risk were duration of diabetes ≥4.5 years, 4 points; total bilirubin ＜6.65 mol/L, 2 points; apoA≥1.18 g/L, 2 points; blood urea≥6.46 mmol/L, 1 points; HbA1c ≥7.75%, 2 points; HDL-c＜1.38 mmol/L, 2 points; diabetic nephropathy, 3 points; fibrinogen, 1 point. The area under the receiver operating characteristic curve was 0.787. The logistic regression analysis showed that the risk factors independently associated with DR were duration of diabetes (β=1.272, OR=3.569, 95%CI 2.283?5.578, P＜0.001), TB (β=0.744, OR=2.104, 95%CI 1.404?3.152, P＜0.001, BUN (β=0.401, OR=1.494, 95%CI 0.996?2.240, P=0.052), HbA1c (β=0.545, OR=1.724, 95%CI 1.165?2.55, P=0.006), HDL-c (β=0.666, OR=1.986, 95%CI 1.149?3.298, P=0.013), diabetic nephropathy (β=1.151, OR=3.162, 95%CI 2.080?4.806, P=0.013), Fg (β=0.333, OR=1.396, 95%CI 0.945?2.061, P=0.094). The risk model was P=1/[1+exp?(?3.799+1.272X1+0.744X2+0.769X3+0.401X4+0.545X5+0.666X6+1.151X7+0.333X8)]. X1= duration of diabetes, X2=TB, X3=apoA, X4=BUN, X5=HbA1c, X6=HDL-c, X7=diabetic nephropathy, X8=Fg. The area under the ROC curve was 0.787 and the Hosmer-Lemeshow test suggested excellent agreement (χ2=10.125, df=8, P=0.256) in model group. The area under the ROC curve was 0.869 and the Hosmer-Lemeshow test suggested excellent agreement (χ2=5.345, df=7, P=0.618) in model group.ConclusionThe area under the ROC curve for DR was 0.787. The duration of diabetes, TB, BUN, HbA1c, HDL-c, diabetic nephropathy, apoA, Fg are the risk factors of DR in T2DM patients.
The valid results of Meta-analysis on biomedical data, will have an important value to clinical practice and health policy making. To review the validity of meta-analysis results, one should consider the following issues: The coverage ratio of included studies, quality of data, publication bias and its effect, heterogeneity, the correct selection of statistical methods as well as clinical significance and external validity of overall effect size. The results of Meta-analysis will keep on updating as new related studies are located and included.
Objective To establish a risk prediction model of diabetic retinopathy (DR) for type 2 diabetic patients (T2DM). Methods A total of 315 T2DM patients (600 eyes) were enrolled in the study. There were 132 males (264 eyes) and 183 females (366 eyes). The mean age was (67.28±12.17) years and the mean diabetes duration was (10.86±7.81) years. The subjects were randomly assigned to model group and check group, each had 252 patients (504 eyes) and 63 patients (126 eyes) respectively. Some basic information including gender, age, education degree and diabetes duration were collected. The probable risk factors of DR including height, weight, blood pressure, fasting glucose, glycosylated hemoglobin (HbA1c), blood urea, serum creatinine, uric acid, triglyceride, total cholesterol, high-density lipoprotein, low density lipoprotein cholesterol and urinary protein. The fundus photograph and the axial length were measured. Multivariate logistic regression was used to analyze the correlative factors of DR and establish the regression equation (risk model). Receiver operating characteristic (ROC) curves were used to determine the cut-off point for the score. The maximum Youden Index was used to determine the threshold of the equation. The check group was used to check the feasibility of the predictive model. Results Among 504 eyes in the model group, 170 eyes were DR and 334 eyes were not. Among 126 eyes in the check group, 45 eyes were DR and 81 eyes were not. Multivariate logistic regression analysis revealed that axial length [β=–0.196, odds ratio (OR)=0.822,P＜0.001], age (β=-0.079,OR=0.924,P＜0.001), diabetes duration (β=0.048,OR=1.049,P=0.001), HbA1c (β=0.184,OR=1.202,P=0.020), urinary protein (β=1.298,OR=3.661,P＜0.001) were correlated with DR significantly and the simplified calculation of the score of DR were as follows:P=7.018–0.196X1–0.079X2+0.048X3+0.148X4+1.298X5 (X1= axial length, X2=age, X3=diabetes duration, X4=glycosylated hemoglobin, X5= urinary protein). The area under the ROC curve for the score DR was 0.800 and the cut-off point of the score was -1.485. The elements of the check group were substituted into the equation to calculate the scores and the scores were compared with the diagnostic threshold to ensure the patients in high-risk of DR. The result of the score showed 84% sensitivity and 59% specificity. ROC curve for the score to predict DR was 0.756. Conclusion Axial length, age, diabetes duration, HbA1c and urinary protein have significant correlation with DR. The sensitivity and specificity of the risk model to predict DR are 84.0% and 59.0% respectively. The area under the ROC curve was 0.756.
Statistical analysis of clinical trials has traditionally relied on frequentist methods, but Bayesian statistics has attracted considerable attention from regulators and researchers in recent years due to its unique advantages, and its use in clinical trials is increasing. Despite the obvious advantages of Bayesian statistics, the complexity of its design, implementation and analysis poses a number of challenges to its practical application, which may lead to an increased risk of unregulated use. This study aims to comprehensively sort out the application scenarios, common methods, special considerations and key elements of reporting of Bayesian statistical methods in clinical trials, with the aim of providing researchers with references for conducting Bayesian clinical trials, and promoting the scientific and rational application of Bayesian statistical methods in clinical trials.
Objective To assess the completion of the under 5 mortality rate (U5MR) of Millennium Development Goals in 194 member countries of WHO, and to analyze the present situation of the global U5MR. Methods Based on the U5MR and the proportion of main causes of death in the "World Health Statistics 2015", the Millennium Development Goals of the decline of U5MR from 1990 to 2013 was assessed, the U5MR was analyzed by comparison between 2000 and 2013. Bivariate Pearson correlation analysis was used to determine the correlation between mortality and the ratio of infection to non infectious diseases and GDP per person in U5MR. Results By 2013, in 194 WHO member states, the U5MR in 46 (23.71%) countries achieved the millennium development goals. Comparison between 2000 and 2013, there was significant difference between low and high mortality groups in six continents (P<0.05), there was no significant difference between the moderate death groups (P>0.05), there was no significant difference in the ratio of infection to non infectious diseases between the middle and low mortality groups (P>0.05), however there was significant difference between the high mortality groups (P<0.05). There was significant difference in the average decline of U5MR and the ratio of non infectious diseases between low and medium, middle and high mortality groups (P<0.05). The Global U5MR had significant regional differences, the highest U5MR was in Africa, the lowest U5MR was in Europe, the medium U5MR was in North America, Oceania, South America, Asia was becoming the middle level. The U5MR was highly correlated with the ratio of infection to non-infectious diseases in every country (r2000y=0.934,r2013y=0.911,P<0.05), and it was low negatively correlated with GDP per capita (r2000y=–0.443,r2013y=–0.433,P<0.05). Conclusions There is a long way to reduce global child mortality. Prevention and control should focus on Africa and Asia. Prevention and control of infectious diseases is an effective measure for middle and high mortality countries. Prevention and control of non-infectious diseases is an important measure for low mortality countries. Increasing health investment is an important means to further reduce global U5MR.
Objective To investigate the pathological mechanism of epileptic comorbid sleep disorder by analyzing the changes of cerebral white matter diffusion tensor in patients with sleep disorder with negative magnetic resonance imaging (MRI) epilepsy based on the method of tract-based spatial statistics (TBSS). Methods MRI negative epilepsy patients comorbid sleep disorder who were epileptic patients treated l in China-Japan Union Hospital of Jilin University from January 2020 to December 2022 completed the Epworth sleepiness scale (ESS) and Pittsburgh sleep quality index (PSQI) tests, and those who complained of sleep disorder and PSQI index ≥11 were monitored by nighttime polysomnography (PSG) and those with objective sleep disorder confirmed by PSG were included in the epilepsy comorbid sleep disorder group. Healthy volunteers with matching gender, age, education were included in the health control group. Diffusion tensor image ( DTI) was collected for all subjects by using a 3.0T magnetic resonance scanner. Diffusion parameters were compared between the two groups using TBSS. Results This study included 36 epilepsy patients comorbid sleep disorder and 35 healthy volunteers. epilepsy patients comorbid sleep disorder showed significantly lower fraction anisotropy (FA) (P<0.05) and significantly higher mean diffusivity (MD) (P<0.05) than the health control group . Brain regions with statistical differences in FA reduction included middle peduncle of cerebellum, genu of corpus callosum, body of corpus callosum, splenium of corpus callosum, anterior corona radiata, external capsule and right posterior thalamic radiation.Brain regions with statistical differences in MD degradation included genu of corpus callosum, body of corpus callosum, anterior limb of internal capsule, anterior corona radiata, superior corona radiata, external capsule and right posterior limb of internal capsul. Conclusion Patients with epilepsy comorbidities with sleep disorders have widespread and symmetric white matter damage.The white matter damage is concentrated in the front of the brain.
The phase-locking relationship between the firings of neuronal action potentials (i.e., spikes) and the oscillations of local field potentials (LFP) reflects important neural coding information. However, the present analysis methods can only determine whether there has phase-locking, but not the different strengths among various types of phase-locking. In the present paper, we used spike-triggered average (STA) signals and the percentage ratio (named φ) of the STA power to the power of original LFP as an index to evaluate the strengths of phase-locking. Experimental recordings obtained from rat hippocampal CA1 region as well as simulation data were used to evaluate the method. The results showed that the index φ changed monotonically as a function of the strength of phase-locking, and it could provide an effective critical value to divide phase-locking from non-phase-locking. Because the calculation of the index does not need pre-filtering, it can avoid the unwanted influences caused by intentionally limiting the frequencies of LFP oscillations such as in the traditional bin statistical method. Therefore, the index φ provides a novel method to investigate the mechanisms underlying neuronal coding in brain.