| Objectives:In order to improve the current situation of conducting congenital heart disease(CHD)epidemiological on-site screening in various regions,it is time-consuming,labor-intensive,consumes too much medical resources,and is fragmented.To explore a new CHD screening model and preliminarily demonstrate the feasibility of Thinklabs one、Iron Man and the remote CHD screening system based on "Iron Man”in CHD screeningMethods:Section one:Epidemiological field screening methods is adopted.From March 26 to 30,2018,and from April 16 to 19,2018,a total of 24932 children aged 4 to 14 from 123 kindergartens and primary schools in 9 townships in Lvchun county,Honghe prefecture,were screened for congenital heart disease.CHD was diagnosed by"cardiac auscultation+echocardiography".Section two:The subjects of the study were selected from the patients who planned to undergo cardiac ultrasound examination in the First Affiliated Hospital of Kunming Medical University and Fuwai Yunnan Cardiovascular Hospital.After consulting the literature,112 CHD patients and 53 non-CHD patients were included according to the formula according to the sensitivity and specificity by referring to the citation literature..Echocardiography was the "gold standard" for the diagnosis of CHD.Finally,135 CHD patients and 57 non-CHD patients were included.The heart sounds were collected by a special person using Thinklabs one and Iron Man,and were stored in the mobile terminal.According to the double-blind design principle,3 senior experts from the department of cardiovascular medicine conducted auscultation on the collected heart sounds and recorded the results of auscultation.The heart sound recorded by Iron Man was uploaded to the network platform and compared with the database to analyze the heart sound,and the results were recorded.To explore the effectiveness of using the heart sound acquisition function of Thinklabs one and Iron Man devices in remote cardiac auscultation.Meanwhile,the computer automatic analysis of remote auscultation by experts and the network analysis system built by expert auscultation and computer automatic analysis were analyzed to analyze the situation of the heart sound collected by Iron Man in the diagnosis of simple CHD.In the process of expert auscultation and data analysis,only the patient’s name and heart sound data as well as the location information of heart sound were provided,while the rest of the patient information was not clear to the analysts.SPSS 19.0 software package was used for statistical analysis by professional statisticians.Results:Section one:1.Among the 24,932 children screened,1122 children with suspicious CHD were found by physical examination(mainly cardiac auscultation)at the initial screening,and 52 children are finally confirmed by cardiac ultrasound,with a prevalence rate of 2.086‰.There is no statistical difference in the prevalence rate among different townships(P>0.05).2.In Lvchun County,the prevalence rate was 3.050‰ among females and 1.263‰ among males,with statistical difference between different genders(P<0.05).3.The prevalence of<1000m at altitude is 1.757‰(6/3414),and that of 1000m at altitude is 2.138‰.4.Among all children with CHD,48.08%are Atrial Septal Defect(ASD),30.77%are Ventricular Septal Defect(VSD),17.31%are Arterial Duct(Patent Ductus Arteriosus,PDA),Tetralogy ofFalos and VSD with PDA accounted for 1.92%,respectively.Section two:1.The heart sounds collected by Iron Man are detected by expert remote auscultation,and the sensitivity,specificity and total coincidence rate of CHD diagnosis are 84.44%,71.93%and80.73%,area under ROC curve is 0.782.The heart sounds collected by Thinklabs one are diagnosed by expert auscultation,and the sensitivity,specificity and total coincidence rate of CHD are 85.93%,71.93%and 81.77%,area under ROC curve is 0.789.There are no statistical difference in the sensitivity,specificity and coincidence rate and area under ROC curve of the heart sounds collected by the two devices(P>0.05).2、The heart sounds collected by Iron Man are analyzed.The sensitivity of computer automatic analysis to diagnose CHD is 89.63%,the specificity is 73.68%,the false negative rate is10.37%,the false positive rate is 26.32%,the Youden’s index is 0.63,and the total coincidence rate is 84.90%,the area under ROC curve is 0.817,the positive predictive value is 88.97%,and the negative predictive value is 75.00%.The sensitivity of the remote CHD screening system to diagnose CHD is 92.59%,the specificity is 61.40%,the false negative rate is 7.41%,the false positive rate is38.60%,the Youden’s index is 0.54,and the total coincidence rate is 83.33%,the area under ROC curve is 0.770,the positive predictive value is 85.03%,and the negative predictive value is 77.78%.Analysis of experts auscultation group network remote screening system,the sensitivity of three groups of specific degree of coincidence of Mc-Nemar inspection,respectively to adjust the inspection level a’=0.05/3=0.017.The sensitivity of remote CHD screening system to diagnose CHD was higher than that of expert auscultation(92.59%vs 84.44%,P=0.001<0.017),the specificity of computer automatic diagnosis of CHD was higher than that of remote CHD screening system(73.68%vs 61.40%,P=0.016<(0.017),showing statistical difference;The rest were not statistically significant.3、By analyzing the heart sounds collected by Iron Man,the accuracy rate of expert auscultation in the diagnosis of ASD was 72.22%,the accuracy rate of remote CHD screening system in the diagnosis of ASD was 86.11%,and the accuracy rate of computer automatic diagnosis was in the middle.Through comparison between groups,the accuracy rate of remote screening system in diagnosing ASD was higher than that of expert auscultation,with statistical difference(86.11%vs 72.22%,P=0.002<0.017),and the rest had no statistical difference.Except that 1 case was missed by expert auscultation in the diagnosis of PDA,with an accuracy rate of 95.24%and the remaining accuracy rate of 100.00%;4、The experts judged the abnormal rate of heart sounds and the diagnostic rate of positive cases through the auscultation of the heart sounds collected by the two devices,and the results were highly consistent,with K values of 0.7080.832(P<0.001).According to the analysis and comparison of the auscultation results in different auscultation areas,the results of the five auscultation areas judged by the experts were consistent,with K value of 0.669-0.785(P<0.001).However,the heart sound abnormality rate of heart sound collected by Iron Man in the auscultation area of aortic valve was higher than that of Thinklabs one(50.00%vs 43.75,P<0.05).5、When Iron Man heart sound was analyzed,the rate of positive cases diagnosed by computer automatic analysis was higher than that by expert auscultation,showing a statistical difference(74.48%vs 67.71%,P=0.041<0.05).There was a moderate consistency between expert auscultation and automated computer analysis in the diagnosis rate of positive cases.Conclusions:1、The prevalence of CHD in school children aged 4 to 14 in Lvchun County is 2.086‰.The prevalence of females is higher than that of males,and the type is mainly ASD.(Altitude range)There is no difference in the prevalence of CHD between different altitudes.2、The accuracy of heart sound acquisition+remote auscultation for CHD diagnosis by Iron Man is similar to that by Thinklabs One,both of which have high sensitivity coincidence rate,but poor specificity.3、The authenticity and reliability of computer automatic analysis of PCG diagnosis of CHD may be comparable to that of expert auscultation;4、The remote CHD screening system combined with expert auscultation and computer automatic analysis may be more sensitive than expert auscultation in diagnosing CHD. |
PDF Full Text Request