應用模糊推論方法於乳癌預測系統設計之研究

Developing the Breast Cancer Prediction System Based on Fuzzy Inference Method

陳振東(Chen-Tung Chen) ； 游惠君(Hui-Chun Yu) ； 陳筑卉(Chu-Hui Chen) ； 黃馨儀(Hsin-Yi Huang)

《創新與經營管理學刊》 9卷2期 (2020/12) Pp. 22-38

https://doi.org/10.6270/JIBM.202012_9(2).0003

摘要

近年來，乳癌已成為我國婦女容易罹患的疾病之一，且乳癌的死亡率也逐年攀升。因此，我國婦女應重視乳癌的篩檢，才能及早發現並有效治療。由於影響乳癌篩檢及判定的因素甚多，且醫事人員判斷的方式及結果往往存在著模糊性與不確定性，而不易在短時間內做出正確的判斷。因此，本研究依據重要特徵屬性的特性建立語意的隸屬函數及規則庫，利用模糊推論方法及解模糊化方法，進行乳癌預測推論系統的設計，以協助醫事人員快速判斷患者罹患乳癌的可能性。資料分析結果顯示，本研究預測系統的平均準確率高達98.5％，且有效降低屬性個數約為56％。因此，本研究所開發的模糊推論預測系統不僅具備學理基礎且有良好的預測效能，能提供重要的參考資訊給醫療人員與機構以進行乳癌判斷與預測。

關鍵字

乳癌預測 ； 特徵屬性 ； 語意隸屬函數 ； 模糊推論系統

並列摘要

In recent years, breast cancer is the most common cancer in Taiwanese women, and the mortality is increasing every year. Therefore, women should pay more attention on screening to find the cancer early and treat it effectively. However, there are many factors affecting breast cancer screening and the subjective judgment of medical crews are always fuzzy and uncertain. It’s not easy to judge correctly in a short period of time. Therefore, this study established the membership functions of linguistic terms and rule base based on the characteristics of attributes. In addition, this study built a breast cancer prediction system by using fuzzy inference and defuzzification method to help medical crews quickly diagnose the possibilities of breast cancer of patients. The results show that the average accuracy of the prediction system is 98.5% and reduce 56% number of attributes. Therefore, the fuzzy inference prediction system proposed by this study not only has the theoretical basis but also has effective prediction capability. It can provide the important information for medical institutions to make the suitable predictions.

並列關鍵字

breast cancer prediction ； characteristics of attributes ； membership functions of linguistic terms ； fuzzy inference system