以類神經網路預測病患之骨質密度

近年由於醫學發達人類壽命延長，罹患骨質疏鬆人口直線上升，而骨質疏鬆症帶來最大的影響便是身體各部位的骨折。以骨質密度狀況推估停經後婦女骨質疏鬆症的盛行率約爲30%，從1999-2001年的健保局資料顯示台灣地區50歲以上成人骨質疏鬆症診斷比率約爲男性1.63%，女性11.35%。但是1996-2000年的健保資料發現國人髖骨骨折比率卻爲全華人地區之冠，可見骨質疏鬆實際狀況有可能被低估以及骨質疏鬆症對國人健康影響之重要性。本研究以台北某醫學中心放射線科之骨質疏鬆病患受檢資料一共359筆。並使用類神經網路爲工具進行分類規則的訓練與預測。最後本研究分別設定兩組變數，分別爲連續變數與類別變數，其中連續變數選項爲：身高、體重、年齡；類別變數選項爲：性別、骨質疏鬆家族史、運動習慣、喝酒、抽菸、喝汽水、喝咖啡；最後由類神經網路訓練之結果爲靈敏度約85%、特異性約53%。

關鍵字

骨質疏鬆症；骨質密度；類神經網路

並列摘要

In recent years, due to the medical development and the extensions of human lives, the population of people who suffer from Osteoporosis increases. The most important effect of Osteoporosis is the fractures of all the body parts. To evaluate by the bone density, the popularity of Osteoporosis among menopause women is 30%. From the data of National Health Insurance Center, the diagnosis of people suffering from Osteoporosis of men who are over 50 years old in Taiwan was about 1.63% and 11.35% to women.. However, the data from 1996~2000 shows that the rate of hip fractures in Taiwan is on the top of Asia. As the results, the true circumstances of Osteoporosis could be underestimate and how important it could effect the human's health. The research is based on the data of 359 patients of Osteoporosis in the department of radiology of a medical center in Taipei and using Automated Natural network of Statistica 8 as a tool to do the trainings and predictions among the different groups. At last, the research set two groups of virables. They are continuous variables and categorical variables. The continuous variables inclusive of height, weight, and ages. The categorical variables are the family medical history of Osteoporosis, the habits of exercise, drinking, smoking, and drinking coffee. Finally, from the results of the trainings of Automated Natural network, the sensitity is about 85% and the specificity is about 53%.

並列關鍵字

Osteoporosis ； Bone density ； Automated Natural Network

國際替代計量

以類神經網路預測病患之骨質密度

全文下載

主題瀏覽