運用資料探勘技術建構攝護腺肥大症手術效果預測模型

攝護腺肥大，為男性專屬的疾病，又稱為良性攝護腺增生，是泌尿科相當普遍的疾病。一半以上的男性在六十歲以後會有攝護腺肥大的症狀，而到了八十歲，大約八成會有這方面的困擾。攝護腺肥大雖屬良性而非惡性的疾病，但除了影響患者排尿與睡眠品質之外，更會造成尿路結石，甚至影響腎功能，降低健康與生活品質。在目前人口老化的社會，這樣的疾病是相當值得注意的。攝護腺肥大的治療方法主要有觀察與等待、藥物療法與手術治療三種。對於輕度的病人只要觀察追蹤即可，對於中重度的病人可以給予藥物上的治療，很多病人都可以透過藥物的使用獲得相當的改善。開刀治療主要是適用在藥物效果不佳或是有其他原因的病人身上，主要分為兩大類：經尿道攝護腺切除術（Transurethral resection of prostate，簡稱TURP）與較新的雷射氣化或是切除手術。兩者都是屬於內視鏡手術，病人外表無傷口，安全性也都不錯，唯雷射手術在安全性方面又比TURP更好一些，特別適用在一些高危險族群，以減少失血或是水中毒發生的機率。然而手術安全性雖然很高，但是並非完全零風險，同時也不是所有的病人手術後效果都十分良好，然而因為給付的關係，不少病人是被建議手術處理。在本研究中，我們試著利用資料探勘的技術，建立攝護腺肥大手術效果預測模型，期望可以在術前預期病人術後的效果，知道哪些病人進行手術的效果不佳，應該避免，就臨床而言是相當有幫助的。本研究證明了資料探勘技術的確可以用來建立攝護腺肥大症手術結果之預測模型，同時可以提供比傳統統計方法達到更佳之預測率，在術前可以有效幫助臨床醫師過濾出開刀效果不佳的病人，以避免不必要之手術。

關鍵字

攝護腺肥大；良性攝護腺增生；前列腺肥大；攝護腺切除手術；資料探勘；決策樹；邏輯期迴歸

並列摘要

Benign prostatic hypertrophy (BPH), also called benign prostatic hyperplasia, is a common disease for aging men. More than 50% of the men in their sixties suffer from BPH and this figure rises to around 80% for 80-year-olds. Although BPH is not a malignancy, it has a major impact on urination, quality of sleep, formation of urolithiasis, and renal function, resulting poor health and lower quality of life. The treatments of BPH include watchful waiting, medications, and surgical intervention. Watchful waiting is suitable for the patients with mild symptoms and signs while medications are more appropriate for those with moderate to severe symptoms and signs and are quite effective in most cases. As for the patients response poorly from medications or not suitable for medical treatments for some reasons, two major types of surgical intervention, namely transurethral resection of prostate (TURP) and laser evaporation or enucleation of prostate, are available for remedy. Although both types of surgical operation incur little danger, they are not risk-free and also don’t guarantee good surgical results. Despite of this, surgical intervention is still often advised due to the payment policy of the Bureau of National Health Insurance. In this study, we adopted a data mining technique, decision tree, to establish a model to predict BPH surgical results. Compared with the traditional logistic regression technique, our approach provides higher accuracy in predicting BPH surgical results that will help clinical staffs screen out those patients who are unlikely to benefit from BPH surgical operations.

並列關鍵字

Benign Prostatic Hypertrophy ； Benign Prostatic Hyperplasia ； BPH ； Prostatic Surgery ； Data Mining ； Decision Tree ； Logistic Regression

參考文獻

1. Akman, T., Binbay, M., Tekinarslan, E., Tepeler, A., Akcay, M., Ozgor, F., . . . Muslumanoglu, A. (2012). Effects of bipolar and monopolar transurethral resection of the prostate on urinary and erectile function: a prospective randomized comparative study. BJU Int.

2. Anderson, J. B., Roehrborn, C. G., Schalken, J. A., & Emberton, M. (2001). The progression of benign prostatic hyperplasia: examining the evidence and determining the risk. Eur Urol, 39(4), 390-399.

3. Bachmann, Alexandera, Woo, Henry H.b, & Wyler, Stephena. (2012). Laser prostatectomy of lower urinary tract symptoms due to benign prostate enlargement: a critical review of evidence. Curr Opin Urol, 22(1), 22-33.

4. Barry, M. J., Girman, C. J., O'Leary, M. P., Walker-Corkery, E. S., Binkowitz, B. S., Cockett, A. T., & Guess, H. A. (1995). Using repeated measures of symptom score, uroflowmetry and prostate specific antigen in the clinical management of prostate disease. Benign Prostatic Hyperplasia Treatment Outcomes Study Group. J Urol, 153(1), 99-103.

5. Berkson, J. (1994). Application of the logistic function to bioassay. Journal of the American Statistical Association, 37, 357-365.

國際替代計量

運用資料探勘技術建構攝護腺肥大症手術效果預測模型

未授權

主題瀏覽