電腦輔助系統應用生理指標評估不孕症病人進行試管嬰兒之懷孕率

A Computer-aided Information System for Detecting Indicators to Access infertility patients pregnancy rate of IVF Treatment

陳文華(Wen-Hua,Chen)

指導教授 ： 徐建業 博士

臺北醫學大學/醫學科技學院/醫學資訊研究所/碩士(2012年)

https://doi.org/10.6831/TMU.2012.00136

摘要

在試管嬰兒（In Vitro Fertilization,IVF）療程中影響療程的成功與否雖然有很多因素，但其中最為主要的因素，則是誘導排卵的反應、胚胎品質及胚胎著床因素等，因此所有醫師在病人進入試管嬰兒療程時，希望借助許多生理指標來治療病患並用這些生理指標提前預判療程是否成功及病患的懷孕機率到底有多大，而常用來作為醫師判別的指標有年齡、BMI值、腰臀比、外觀以及許多抽血檢驗的生理數值；最常用的如 濾泡刺激素(Follicle-stimulating—hormone,FSH)、黃體刺激素 (Luteinizing hormone,LH)、泌乳素 (Prolatine,PRL)、睪丸酮(Testosterone)、SHBG (Sex hormone-binding globulin) 以及抗穆氏管荷爾蒙 (Anti-Mullerian hormone,AMH)，運用針對各種不同的生理指標進行研究與測試中，並藉由研究與實驗的過程，了解指標的變化，進而能夠達到提升試管嬰兒療程的懷孕率目的。近年來有科學家研究指出，上述的生理指標對胚胎品質影響甚巨，根據統計分析顯示病患年齡、AMH等指標在臨床上對懷孕率是具有顯著的意義。然而將該兩個指標所建立的預測模型與所有指標建立預測模型比較，確定後者的模型準確率較高，並運用驗證與臨床判斷比較確定可行性後，進而建立電腦系統的預測模組，運用在試管嬰兒療程懷孕與否的預測，對不孕症醫療是具有非常重要的意義。
本研究的目的是針對在進行臺灣不孕症療程的病人，將病人的懷孕率與各項生理指數的相關性做分析，建立試管嬰兒療程懷孕率的預測模型及電腦系統懷孕率預測模組。本研究對象一共有240 名有效樣本，將所有研究對象的生理指數都藉由醫護人員輸入電腦完成後，運用SPSS進一步的統計分析，首先確立生理指標與懷孕率具關聯性的意義及運用指標建立預測模型，並運用人工智慧類神經網路與醫師臨床的判斷對該模型以同樣的樣本做檢驗及比較確認，確認預測模型的可行性，完成對試管嬰兒療程懷孕率的預測提供具體做法。

關鍵字

電腦系統；不孕症；試管嬰兒；預測模型；懷孕率

並列摘要

In (In Vitro Fertilization, IVF) treatment there were many factors affecting the success of it. The most important factors are the response to ovulation induction, embryo quality and embryo implantation. Therefore before patient access to IVF treatment, all physicians wish to predict the possibility of success by physiological indicators. Commonly used as a physician determine indicators were Age,Body Mass Index(BMI) values, Waist to Hip Ratio(W_H), appearance, and many blood tests of the physiological value; the most commonly used, such as Follicle-stimulating hormone(FSH), Luteinizing hormone(LH),Prolatine (PRL),Testosterone(T),Sex hormone-binding globulin (SHBG) and Anti-Mullerian hormone (AMH), Therefore, research and testing for a variety of physiological indicator , is nothing but to study and experiment by the process of understanding the changes indicators and can enhance the pregnancy rates of IVF treatment. But in recent years some scientists have pointed out: the physiological indicators affect embryo quality highly. Which AMH were more significant on clinical pregnancy rate. Therefore, how to use all of the above physiological indicators to generate the prediction Model and thus the establishment of computer information systems, and applied to predict of the IVF pregnancy or not, is a very important issue of infertility treament. On medical information application field, First of all is to build computer systems to provide convenience for physician to import patient data and basic physiological parameters, and then the use of physiological indicators in database to do research and statistical analysis of In Vitro Fertilization(IVF) treatment. And establish the Predict Model base on statistical analysis,and thus provide a predict result to doctors on the pregnancy rates of IVF.
The purpose of this study is aimed at Taiwan IVF patients , the relevance of the computer system detect the physiological indicators of IVF patients and predict their pregnancy rate and establish Predict Model of IVF treatment.
Total of subjects are 240, the physiological indicators of all subjects are completed import by medical staff for more further research and analysis. Firstly, use by SPSS computer statistical software for all samples of all the physiological indicators with BMI, W_H ratio and Age. Associated with pregnancy or not to do statistical analysis, and then get the prediction Model and the pregnancy probability, and test the pregnancy probability by artificialthe wisdom of the neural network software..
Artificial intelligence science since the 1950s, we derived a variety of information technology to assist in the decision-making, which, inspired by neurobiology, neural networks, parallel computing, and highly fault-tolerant features, specializes in the classification of a large number of heterogeneous data. Especially medical treatment due to clinical data collection is not easy and a great variety and difficult to handle in the usual way, is that since 1985, the use of neural network applications in medical decision-making will be turbulent.
According statistical results, Including Age and AMH have significant meaning for pregnant, then the use of logistic regression of SPSS statistical software, to get the formula and the probability of the prediction Model, Using Medcalc statistical software to create the ROC curve. Finally, test all physiological indicators by the neural software, and to get another set of probability values. Make the Validation of ROC curves between obtained probability value the neural software with SPSS. Then compare the Model predicted results with infertility doctors predict , Finally, do sampling in 2012 patients for Prediction and verification. To confirm the feasibility of the prediction Model, Thereby increasing the possibility of the prediction of IVF pregnancy rate.

並列關鍵字

Computer System；Infertility；In Vitro Fertilization(IVF)；Prediction Model；Pregnancy Rate

參考文獻

Andre H. Philippe B. David B, et al .Serum antimullerian hormone/mullerian-inhibiting substance appears to be a more discriminatory marker of assisted reproductive technology outcome than follicle-stimulating hormone, inhibin B, or estradiol. Fertility and Sterility.2004;82(5): 1323-1329.

Cem F. Tayfun K. Elif B, et al. Early follicular antimullerian hormone as an indicator of ovarian reserve. Fertility and Sterility .2006;85(3):592-596

Dorothea M. Nick A. Mengistu Y. Robert K.Martin H. Statistically significant changes of antimullerian hormone and inhibin levels during the physiologic menstrual cycle in reproductive age women. Fertility and Sterility. 2008;89;4:927-933.

Dorothea M. Jean G. Martin H. Nick A. Anti-Mullerian hormone and inhibin B as predictors of pregnancy after treatment by in vitro fertilization/intracytoplasmic sperm injection. Fertility and Sterility.2008;90(6): 2203-2210.

D.Barad, A. Weghofer, N. Gleicher. How predictive of basic pregnancy potential are extremly low levels of anti-mullerian hormone (AMH)?. Fertility and Sterility .2009;92(3):178-179