膽結石病患伴隨腎結石之評估研究

A Study on the Assessment of the Gall Stones Disease Associated with Renal Stone

林振豪(Zhen-Hao Lin)

指導教授 ： 張俊郎

國立虎尾科技大學/管理學院/工業管理系工業工程與管理碩士班/碩士(2017年)

摘要

隨著醫療水準漸漸提高與生活環境改善，臺灣人口結構逐漸邁向高齡化社會甚至超高齡化，導致慢性疾病逐漸滲入。膽結石和腎結石兩者皆是近代常見的公共健康問題，現代人為了工作生活忙碌，導致平時生活作息不正常、缺乏運動，以及三餐不正常加上飲食型態日益西餐化，高熱量、高脂肪、高醣類的東西常常攝取過多，使的膽結石的罹患率逐漸升高。而膽結石與腎結石同為結石類疾病，兩者間也有共同的危險因子，但探討兩者之間關聯性的研究並不多。
　　本研究以某醫療機構資料庫中膽結石病患為研究對象，透過文獻探討與醫師訪談，篩選出會增加腎結石疾病風險之重要因子，運用人工智慧中的粒子群最佳化演算法、基因邏輯斯迴歸演算法、交叉熵演算法計算因子權重值，並分別結合倒傳遞類神經網路與支援向量機，以建構六種預測模型與三種案例式推理系統評估，來評估膽結石患者後續是否產生腎結石的風險。
　　研究結果顯示六種預測模型，經由傅利曼統計檢定，並無顯著差異，皆適合作為本研究之預測模型，其平均測試準確率皆達89%以上ROC曲線下面積也在0.85以上；而三種案例式推理系統模型，雖然以粒子群演算法之模型準確率89.01%與ROC曲線下面積0.894較佳，但檢定結果三種模型亦無顯著差異，皆適合做為案例式推理評估系統之權重計算。

關鍵字

膽結石 ； 腎結石 ； 粒子群最佳化演算法 ； 基因邏輯斯迴歸演算法 ； 交叉熵演算法 ； 倒傳遞類神經網路 ； 支援向量機 ； 案例式推理

並列摘要

With the gradual improvement of medical standards and the improved living environment, Taiwan’s population structure has gradually headed towards the aging society or even ultra-aging, leading to the gradual infiltration of chronic diseases. Gallstones and kidney stones are both today’s common public health problems. Modern people lead a busy lifestyle, resulting in disrupted daily routine, lack of exercise, and failure to take meals regularly. In addition, the increasingly westernized diet pattern and excessive intake of high-calorie, high-fat, and high-sugar foods have resulted in a gradual rise in gallstone incidence rates. Gallstones and kidney stones both fall under the stone disease category and share common risk factors. However, researches exploring the connection between the two remain scarce.
　　In this study, patients with gallstones from the database of an anonymous medical institution were adopted as research participants. Through literature reviews and interviews with physicians, the important factors contributing to increased risk of gallstone disease were screened. Through the use of algorithm in artificial intelligence such as particle swarm optimization algorithm, genetic logistic regression algorithm, and cross entropy algorithm, the factor weights were calculated. The back propagation neural network and support vector machine were conjunctively employed to construct six prediction models and three case-based reasoning systems in order to evaluate whether patients with gallstones are at risk of future gallstones in the future.
　　Research results show that after the Friedman’s test, the six predictive models showed no significant differences and are all suitable as predictive models for research. The models all reached 89% average test accuracy, and the area under the ROC curve was above 0.85. For the three case-based reasoning system models, although the model of particle swarm optimization algorithm with 89.01% accuracy and 0.894 area under the ROC curve was better, the test results of the three models show no significant differences, making them suitable for calculating the weight of the case-based reasoning assessment system.

並列關鍵字

gallstones ； kidney stones ； particle swarm optimization algorithm ； genetic logistic regression algorithm ； cross entropy algorithm ； back propagation neural network ； support vector machine ； case-based reasoning

參考文獻

6.李建邦、俞星辰(2014)，「以粒子群最佳化演算法結合分類器預測疾病急搜尋變數之研究」數據分析，9卷1期，頁57－74。

28.劉晏孜、陳琬青、林益卿(2012)，「腎結石之治療與預防」，家庭醫學與基層醫療，27卷、11期、頁384-390。

13.孫瑛霙，王明庸，藍天雄(2015)，「應用倒傳遞神經網路於負載啟斷開關接觸電阻預測之研究」，管理資訊計算，4卷，1期，頁142-149。

7.李維平 王雅賢 江正文(2008)，「粒子群最佳化演算法改良之研究」，科學與工程技術期刊，第4卷，第2期，頁51-62。

31.劉蓓麗、李宜致等人(2007)，「肥胖病人膽結石生成之預測因子研究」，澄清醫護管理雜誌，3卷，3期，頁29-36。