Font Size: a A A

Data Processing And Analysis Of Thalassemia And Folic Acid Screening By Pre - Pregnancy Disease

Posted on:2016-12-14Degree:MasterType:Thesis
Country:ChinaCandidate:L F LiuFull Text:PDF
GTID:2134330470970470Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Every year in China the rate of birth defects and the rate of congenital disabilities which occur among 0-4 years old infants total 5%, that is to say, annually more than 1 million infants are born with disabilities. Therefore, the pre-pregnancy physical examination is significant for the birth of a healthy infant. From the perspectives of data mining and statics this paper focuses on the thalassaemia test and the folate level test included in the pre-pregnancy physical examination.Aiming at the classification problem of multi-property values in the thalassaemia screening and classifying model, many studies use a manually set threshold to determine the characterizing attribute. This method is subjective so that the precision of classification will be influenced. To solve such a problem a new method is proposed in this paper. First, sort the characterizing attributes with the Bhattacharyya distance; second, use the genetic algorithm to determine the best SVM mode, then screen out the characterizing attributes by traversal to obtain the classification precision; third, choose the threshold point with the highest classification precision. The experiment results show that the method proposed in this paper is better the manual one, and has good performance on the thalassaemia test, especially for processing the large sample threshold.For the study of thalassaemia screening, this paper mainly compares the morbidity statistics between different races and different areas in Yunnan Province in order to realize the relations between thalassaemia and races/areas. In traditional ways the statics of routine blood tests and haemoglobin electrophoresis tests are analysed, but in this paper the characterizing attributes like the gender, age, birthplace and race of both the healthy and ill are added, meanwhile, the best logistic regression and SVM model is built to screen thalassaemia on the basis of the regular classification model for six types of thalassaemia. The experiments show that there is no significant association between the thalassaemia morbidity and race/area in Yunnan Province. However the screen precision and diagnostic OR have increased obviously due to the six-type classification model in comparison with the traditional method.For the study of folate level tests, the RBC folate level of the reproductive population in Yunnan Province is analysed to set up a reference range of the RBC folate level. The relationship between the lack of folate and the history of abnormal pregnancy and parturition is studied by comparing the RBC folate level between the normal people and the people who have the history of abnormal pregnancy and parturition in their reproductive years. The statistics data show that the RBC folate reference level of the 20-29 years old women at child-bearing age is 11.31~35.73 ng/ml, and the one of the 20-29 years old men is 10.85~33.59 ng/ml, the measured value of women is higher than men. Besides, the folate level differences between different races in Yunnan Province are not statistically significant (P>0.05). By analysing the 2639 samples in the normal group and 246 samples in the group of people with abnormal pregnancy and parturition history, the difference between the folate level of the normal people and the people who have ever given birth to children with encephalopathy is statistically significant (P<0.05). By comparing the folate level of the normal group with the people with history of fetal malformation, spontaneous abortion, embryo damage, induced abortion caused by neural tube defects and birth history of infants with Down’s syndrome and other chromosomal abnormalities, the difference is not statistically significant (P>0.05).The algorithm which automatically finds the best threshold can better determine the best threshold point of the characterizing attributes, so better clinic screening results can be obtained by the optimal classification model. The determination of RBC folate level reference range in Yunnan provides theoretical evidence for the folate supplementation of people at reproductive age in Yunnan. Folate supplementation is meaningful for preventing birth defects, which should be encouraged among the reproductive population.
Keywords/Search Tags:Pre-pregnancy check, Data mining, Threshold, Thalassemia, Classification model, Folate
PDF Full Text Request
Related items