基于卷积神经网络的肠道管理评估研究

Intestinal Management and Assessment Based on Convolutional Neural Network

摘要: 近年来，肠道健康问题逐渐呈现年轻化趋势，便秘、腹胀、消化不良等肠道困扰现象在青年和中年群体中频繁出现。在肠道疾病的发病过程中，人体存在一段自我调节和恢复的时期，这一时期对于预防肠道疾病的发生至关重要。然而，由于缺乏及时提醒，普通人往往难以察觉这一阶段的生理变化，从而未能及时采取措施进行身体调理和预防肠道疾病的发生。因此，本研究提出了适用于肠道管理评估的卷积神经网络算法(Convolutional Neural Network, CNN)，将加权k最近邻算法(k-Nearest Neighbors, KNN)、支持向量机(Support Vector Machine, SVM)做对照组并对照医生的诊断结果，得出该评估系统算法的准确性。结果显示，加权KNN的准确率为75.33%，SVM的准确率为88.00%，CNN的准确率为94.6%。说明基于CNN的肠道管理评估算法研究可以较为精准地估计肠道健康等级，具有较好的应用前景。

Abstract: In recent years, intestinal health problems have gradually shown a trend of rejuvenation, and intestinal disturbances such as constipation, bloating, and dyspepsia are frequently seen in young and middle-aged groups. During the onset of intestinal diseases, there exists a period of self-regulation and recovery in the human body, which is crucial for the prevention of intestinal diseases. However, due to the lack of timely reminders, it is often difficult for the general population to detect physiological changes during this phase, thus failing to take timely measures for body conditioning and prevention of intestinal diseases. Therefore, this study proposes a Convolutional Neural Network (CNN) algorithm suitable for the assessment of intestinal management, and the weighted k-Nearest Neighbour algorithm (KNN), Support Vector Machines (SVM) are used as a control group and against the diagnosis of doctors to derive the accuracy of the algorithm of this assessment system. The results show that the accuracy of weighted KNN is 75.33%, SVM is 88.00% and CNN is 94.6%. It shows that the research on the assessment algorithm of intestinal management based on CNN can estimate the intestinal health grade more accurately, which has a good application prospect.