IVF(In vitro fertilization)中譯為試管嬰兒治療。將卵子與精子取出,在人為的操作下進行體外受精,培養成胚胎,再將胚胎植回母體,試管內進行整個過程只有2-6天內,完整的過程稱為試管嬰兒。對於不孕症治療,是一項有效的醫療技術,對於每位患者,最在意的就是成功與否,依目前的技術仍無法百分百的成功。初期醫師會根據病患的年齡、不孕症的病因、濾泡刺激素(follicle stimulation hormone)指數,預測患者大概的成功機率,在進行療程中,會依夫妻的精蟲之數量與活動力、卵巢功能、子宮內膜對胚胎之接受度等等的情形,再給予病人不同的意見與療程,提高病人的成功率。本研究目的是應用資料探勘(data mining)技術於IVF的資料分析,建構出IVF的結果預測模式,找出IVF失敗與成功的規則。為了提高資料探勘的結果,本研究使用隨機森林(Random Forest)得出最高分類72%,也可以用決策樹C4.5與RIPPER 演算法得出部分規則模型。本研究結果,可提供醫師對IVF患者的成功率預測之參考。
The Purpose of this study was to investigate IVF (In Vitro Fertiliztion), Scientific name is la fertilization in vitro, people alwas call test tube baby. It’s a good medical tedchnology for infertility. Every patient who most care is success or failure. Exsiting techonology still can’t make it 100% success. The first course of treatment, phtsician will ask patient age, pathogen, follicle-stimulating hormone indes to predicating success rate. Man and wife’s treatment will depend on sperm count and healty or not, oarium activity, endometrium fory embryo acceptance etc. Give patients opnion and treatment to promote auccess rate. This research is using dataming techmology to predict IVF and construct IVF model. Find rule about success and failaur rule .This study using Random Forest have best predicting 72%. Also can use Decision tree C4.5 and RIPPER algorithm, can construct model and rule. The study provided a physician on the success rate of ivf patient predictive reference.