| Summary: | 碩士 === 國立屏東科技大學 === 資訊管理系所 === 101 === Currently the gold immunochromatography was used to do prenatal screening in remote inland area where these two diseases frequently occur. However, this method always gets more pregnant women of false-positive and need amniocentesis confirm. This is a big burden for hospitals and pregnant women. This study established a set of intelligent prenatal detection system of Down syndrome and neural tube defects to improve the situation. (1) Historical data acquisition rule base module - analyze historical data and establish a risk ratio rule base by using data mining technology and self-organizing map neural network (Self-Organizing Map; SOM). (2) Detection analysis module - this system automatically analyzes and matches the rule base to assess onset risk of Down syndrome and neural tube defect in new subjects. (3) Intelligent query function module - the relevant information of pregnant woman will be screened by using data mining technology. This system not only provides timely the required data for physicians, but also provides query and analysis of batch results of pregnant women. Meanwhile, the system can feedback test results to correct risk ration rule base. The gold immunochromatography is currently used to do prenatal screening in hospital and the false positive rate of Down syndrome and neural tube defects is 3.5% and 1.5% .Under the premise in the absence of missing positive patients, this system can effectively reduce the false positive rate of two kinds of illness by a random sampling test. The false positive rate of Down syndrome was reduced by an average 20.6%, the false positive rate of neural tube defects was reduced by an average 54.9%. Meanwhile, the system not only provides the basic data of pregnant women for physicians, but also auto filters and prompts reexamination function for pregnant women with high risk. This system can substantially reduce the false positive number of high-risk pregnancies and reduce the pressure on hospitals and pregnant women.
|
|---|
