強化粗糙集應用於闌尾炎之分類

Refining rough set for applying classification of appendicitis

決策科學應用在醫學相關領域的應用近十年來發展的相當迅速，醫學疾病的診斷，各項檢查值與疾病特徵的判斷常與人類的思考模式相似，具有不精確的特性，有時是屬性介在二種不同的疾病時仍然難以區別，而它具有近似性、邊界，基於此一特性，應用粗糙集合理論於醫學領域之探勘具有相當良好的效果。闌尾炎為急診醫學科常見之急性發作疾病，它需要綜合每一與此疾病相關的屬性來判斷，而此一腹痛疾病好發之年齡並不特定，急診醫生在針對患者自訴之抱怨有時亦即難時判斷出是否為闌尾炎，而除了生理上之明顯特徵，如發燒、腹部腫脹、腹痛期間之掌握，各種檢查檢驗如血液、尿液、腹部X光等都具有參考價值。粗糙集合對於多屬性疾病具有相當良好的分類能力，本研究在區別五種類別疾病獲得了 91.1％之正確率，在分類個別疾病上有些亦高達100％，而萃取之規則之涵蓋率也達95.2％，可做為後發展其它疾病如胸痛、頭痛等相關疾病之鑑別的模式。

The decision science is applied in medicine, which is developing speedy since this decade. Rough set is innovation for classifying in medicine, which was proposed by Pawlak in 1980s, which used to classification and prediction. The appendicitis is often seen in emergent room, but to discriminate appendicitis feature from abdominal pain diseases, which not only depend on complain of patient, but also need more information such as KUB (Kidney, Ureter, and Bladder), Culture, Urine, and so on. Appendicitis diagnosis is ambiguous same as the thinking of human under uncertain, imprecision. For conquer the ambiguous by reason the approximation and boundary, we employ rough set theory and PCA aim at the appendicitis to classify. This paper achieves the 91.1% total accuracy of classification from 5 abdominal pain diseases and 95.2% of coverage of extracted rules.