一個新的糖尿病患者飲食推薦方法

A New Dietary Recommendations Method for Diabetic

糖尿病已成為台灣的文明病之一，高熱量、高脂肪、高蛋白的食物更使得國內人罹患糖尿病的人數逐年增加。而患者雖然能通過諮詢營養師的方式獲得飲食資訊，但礙於國內營養師的人數不足且本身專業不同並非都是糖尿病專長的營養師。也因此在本研究希望建立一個專為糖尿病患者的飲食推薦系統。本研究使用Protégé來輔助建立我們的食品知識庫。並使用Fuzzy Logic做為系統之前導推論，依患者之身體資訊推論出適合使用者每日所需之總熱量值，並使用JENA做為我們的推論器並建立我們的知識規則。最後，我們將配合營養師對糖尿病飲食的專業，結合類背包演算法與TOPSIS演算法對推薦出的飲食組合進行進一步的排序以求更科學合理的推薦。

With the rapid development of economy, the style of diet has changed a lot. Due to high calorie, high fat and high protein foods, lot of people have diabetes problem and the diabetes patients increase quickly each year. In this paper, we will propose a diet recommendation system. Firstly food ontology is established using the tool of Protégé. Next, fuzzy Logic system combined with Jena inference to infer daily calories demand for patients, according to the patient's health information. Then, Knapsack-like algorithm is used to figure out diet combinations. Finally, TOPSIS algorithm is applied to sort the diet combinations based on the guiding of the nutritionist.