Font Size: a A A

Design And Experimental Analysis Of Jaundice Detection System Based On Double Detectors

Posted on:2020-02-11Degree:MasterType:Thesis
Country:ChinaCandidate:S J WuFull Text:PDF
GTID:2392330599962021Subject:Instrument Science and Technology
Abstract/Summary:PDF Full Text Request
As one of the common diseases in newborns,neonatal jaundice happens due to abnormal bilirubin metabolism,which leads to the increase of bilirubin concentration.In the severest situation,jaundice can cause kernicterus.Kernicterus is able to result in the irreversible damage to the nervous system,and even child death.Therefore,detecting the concentration of bilirubin is of great significance to prevent and treat bilirubin encephalopathy.At present,when jaundice is detected with specific equipment,its accuracy is affected by the developmental differences of newborns and the range of bilirubin concentration.So,detection cannot be done in a proper and actual time.This paper proposes a method of detection based on double detectors.Firstly,to analysis the theory of the detection and make an improvement of single detection already carried out,theoretical model by the double detector is established;Secondly,to propose the overall method of design,the design structure of the detection by double detector,hardware and software in the system are completed;Finally,to obtain the signal of bilirubin concentration and its eigenvalue by amount of experiments,,do the regression analysis of the eigenvalue by referring to bilirubin concentration,and to overcome the measurement tolerance introduced by the method itself and improve the accuracy of model,the prediction model of bilirubin concentration is constructed,which is based on MLR-BP neural network method.The correlation coefficient of the model was 0.996,and the square root of detection tolerance was 5.196?mol/L.Finally,do the research and analyze the system of jaundice by double detection.The MLR-BP neural network was used to establish a single detector bilirubin concentration model,and 30 groups of premature newborn with neonatal jaundice were randomly selected for prediction.The square root of bilirubin by predicting single detection model was11.74 ?mol/L.The square root by predicting double detection model was only 6.11 ?mol/L.According to data of treating neonatal jaundice in the research and analysis of ROC curve line,the accuracy of neonatal jaundice in the double detector was 97%,and the area under the ROC curve line was 0.909.The results showed that the system of double detector to jaundice can overcome the detection tolerance caused by the different developmental degree of the newborns and improve the accuracy of detection and treatment of neonatal jaundice in newborns.
Keywords/Search Tags:Neonatal jaundice, Dual detector jaundice detection, MLR-BP neural network combination method, Experiment analysis
PDF Full Text Request
Related items