台灣人工生殖 (In vitro fertilization, IVF) 技術享譽國際醫學界,隨著現代人生活與工作壓力的增加,不孕夫妻的比率越來越高,求助醫療所之不孕症患者日益增加。人工生殖試管嬰兒的整個流程步驟,包括病患的基本資料建檔、病患的藥物注射給予、開刀房手術取卵、受精、胚胎的培養、胚胎的記錄、胚胎的植入、多餘胚胎的冷凍保存、冷凍胚胎的植入、懷孕的最終結果(流產、子宮外孕或足月生產等)紀錄,及實驗室的品質管制 (如CO2培養箱、液態氮桶監內液態氮監控、培養液PH值監控) 等階段,由於內容相當複雜,而且每一細部的環節皆會影響到懷孕率最終成果。 然而試管嬰兒的療程費用是相當昂貴的,且受療者須忍受身心的考驗也易引發一些副作用如:出血、感染、膀胱穿刺等。而且成功懷孕因素眾多,醫師雖可以依病人年齡、卵泡刺激素(Folliclestimulating hormone)及不孕症的診斷等臨床經驗來決定成功懷孕機率的大小,但有許多其他因素也會影響IVF的成功率,如卵子及精子的品質、植入胚胎的數量及是否有子宮卵巢的病變等,醫師卻很難針對每一組IVF受療者擬定個別的治療決策。本研究採用資料探勘(Data mining, DM)的技術,結合粒子群最佳化方法與蜜蜂交配演算法等技術分析IVF資料庫,探討影響IVF成功或失敗主因及成功因素的規則,建構一套預測模式。研究結果以期能提供給醫師及不孕症患者相關診斷時的決策參考。
The technique for in vitro fertilization(IVF) in Taiwan has been one of the best medical practices in developed countries. With increasing stress in work environment and daily life, more couples have encountered infertility issues and are seeking medical treatments from fertility clinics in hospitals. The procedure of IVF includes data collection for patients’ basic medical history, ovarian stimulating medication injection, surgically egg extraction, united eggs and sperms, fertilization of embryos, embryo transfer, cryopreservation for embryos, frozen embryos transfer, data of pregnancy results (for example, miscarriage, ectopic pregnancy, child birth), and quality control in the laboratory (such as control of CO2 incubator, liquid nitrogen, PH value in solution). The purpose of this project is to construct a decision-making support model for success rate prediction of IVF. We proposed a prediction model which utilized Particle Swarm Optimization algorithm and Honey Bees Optimization algorithm to analysis IVF database. The key factors of successful IVF are analyzed. Then, the decision tree model is used to construct the prediction rule model. The results provide physicians and patients with proper decision support when doing the in vitro fertilization treatment.