頁籤選單縮合
題名 | The Trial of C4.5/ID3 in Predicting the Presence of Rectosigmoid Polyps: A Preliminary Report in National Taiwan University Hospital |
---|---|
作者 | Chiang,I-jen; Hsu,Jane Yung-jen; Lin,Win-li; Wong,Jau-min; Shieh,Ming-jium; Wang,Cheng-yi; | 書刊名 | 醫學工程 |
卷期 | 8:2 1996.04[民85.04] |
頁次 | 頁18-23 |
分類號 | 416.245 |
關鍵詞 | C4.5/ID3; Machine learning; Polyp; Sigmoidoscopy; |
語文 | 英文(English) |
英文摘要 | C4.5/ID3 is a well-known machine learning methodology in some domains of industrial manufacturing and business. It works on the basis of given attribute-value data to properly discriminate important classification attributes among groups. In the present study, we applied this learning method to routine physical checkup data to determine the predictive value for colorectal polyps in the rectosigmoid region in a general population, comparing data with signmoidoscopic findings. Filtering only the patients who had undergone 60 cm sigmoidosocopy, we selected 29 personal and biochemical attributes: blood type, sex, age, serum cholesterol, triglyceride, total protein, albumin/globulin (A/G), albumin, Zinc Turbit Test, direct bilirubin, total bilirubin, alkaline phosphotatase, acid phosphotatase, alanine aminotransferase, aspirate aminotransferase, hemogram of white blood cell, red blood cells, mean corpuscle volume, hemoglobin, hematocrit and platelets, urinary occult blood, protein and sugar, stool occult blood (day 1), stool occult blood (day 2), hemoglobin AIC, alcohol consumption or not, and smoking. At first, C4.5 was applied to 2746 patients to construct an antecedent-consequence rule for polyp prediction. The antecedent consequence rule for polyp prediction thus constructed was then used on another testing set of 2384 patients chosen the next year. When compared with the findings of the sigmoidoscopy, we found that C4.5/ID3 failed to identify any significant risk factor for rectosigmoid polyps among the 29 attributes filtered. We conclude that so far, computer simulation study attempting to combine putative risk factors in the prediction of colorectal polyps in general populations is immature; this implies that neither single etiological factor nor any combination is likely to be applicable as a basis for screening at present. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。