疾病檢測及風險評估決策輔助系統建立之研究－以糖尿病為研究對象

A Decision Support System for Disease Diagnosis and Risk Evaluation-With Diabetes as the Research Object

邱瑞科(Ruey-Kei Chiu) ； 黃珈瑩(Jia-Ying Huang) ； 張彥群(Yan-Qun Zhang)

《醫療資訊雜誌》 19卷3期 (2010/09) Pp. 1-22

https://doi.org/10.29829/TJTAMI.201009.0001

摘要

我國糖尿病發生率逐年增加，國人健康與此相關的課題，近年來越受到政府衛生主管機關及醫療機構的關注。而因疾病本身不易檢測且初期症狀不明顯，往往會因不易察覺而被忽略，一旦罹患病人便須接受長期治療與控制，因此若能及早發現並預防，即可有效降低或延緩糖尿病的發生。我們深信藉由電腦爲基決策支援系統的輔助必可更有效協助醫醫療專業人員臨床診斷與檢測。 本研究首先藉由文獻探討及彙整專科醫師的專業知識，以釐清引發糖尿病之顯著危險因子，並運用人工智慧技術之類神經網路建立檢測糖尿病與代謝症候群分類之模型，並將分類結果進一步藉由模糊專家系統推論，以判定糖尿病之風險嚴重程度，藉此發展一完整之糖尿病線上檢測暨風險評估輔助系統，提供醫師了解糖尿病患者病情並作爲更精確診斷之決策資訊，並給予早期治療及適當生活調適指引。 經實驗驗證，其結果顯示本研究所發展之系統確實可以達到與專業判斷同樣的水準，且能提供專業醫療人員可迅速地進行更精確診斷與診療之參考資訊來維持並提昇病人的健康。

關鍵字

人工智慧 ； 類神經網路 ； 模糊專家系統 ； 糖尿病 ； 代謝症候群

並列摘要

The incidence of diabetes has increased every year in Taiwan. The issue regarding this disease of civilian health was paid more attention by governmental health authorities and medical institutions in recent year. It is very often overlooked because it is unconsciously found early to the disease. Once contracting the disease, people must suffer from a long-term treatment and control. Therefore, if early detection and prevention can be carried out, it may effectively reduce or avoid the incidence of diabetes. We believe that a computer-based decision support system must be able to effectively support the medical professionals to conduct clinic diagnosis and detection of the diabetics with its level of risk.
The study of this paper, firstly the risk factors of causing diabetes were identified by reviewing literatures and collecting professional knowledge of physicians. Then, the neural network of artificial intelligence technology was applied to conduct the classification of diabetes. Once the diabetes was identified, the fuzzy expert system was then applied to measure the risk level of the diabetes. By means of the aid of this comprehensive on-line detection and risk evaluation system, the information generated from the system may be taken by professional physicians to understand the disease condition and proceed to a more precise diagnosis, whilst provide an early diagnosis and the proper guidelines for living adjustment to a diabetes.
Through the verification of system experiment, the result shows the system developed in this study can reach to an analogous or even better level of diagnostic effect which is provided by professional medical personnel. As a result, the health of a diabetic can thus be more effectively maintained and improved as well.

並列關鍵字

Artificial Intelligence ； Neural Network ； Fuzzy Expert System ； Diabetes ； Metabolic Syndrome