论文部分内容阅读
目的:准确测量血压(BP)对于流行病学研究、筛查规划、调研研究和高血压相关病变(冠心病、中风、肾衰竭等)的早期诊断及预防有重要意义。被测者体位对于准确测量血压有重要影响。血压测量指南建议测试时被测者应在后背有支撑的情况下保持坐姿,以避免血压读数偏高。本文使用混杂模型预测血压对于无支撑后背的反应。创新点:本文考虑血压正常和高血压测试者的人体预测变量(如年龄、身高、体重、体块指数和上臂周长(AC)),使用基于PCA的前向逐步回归(PCA-SWR)、基于PCA的人工神经网络(PCA-ANN)、基于PCA的自适应神经模糊推理系统(PCA-ANFIS)和基于PCA的最小方差支持向量机(PCA-LS-SVM)等模型预测血压对无支撑后背的反应。方法:使用PCA消除人体预测变量间的多重共线性,并在原始数据集中选取主元(PC)。所选主元被输入至所建立预测模型用于建模及测试。结论:通过评估合适的统计指标(确定性系数、平均平方根误差、平均绝对百分比误差),得出较之其他模型,PCA-LS-SVM对于预测血压反应较有前景。此评估也展示了混杂模型在预测生物医学领域其他参数时的重要性和先进性。
OBJECTIVE: Accurate measurement of blood pressure (BP) is of great importance for the early diagnosis and prevention of epidemiological studies, screening programs, research studies and hypertension-related diseases (coronary heart disease, stroke, renal failure, etc.). The subject’s position has an important impact on the accurate measurement of blood pressure. Blood pressure measurement guidelines suggest that subjects should be tested in the back with the support of the situation to maintain the posture, in order to avoid high blood pressure readings. This article uses a hybrid model to predict the response of blood pressure to unsupported back. Innovations: This article considers the predictor variables of normotensive and hypertensive subjects such as age, height, weight, body mass index, and upper arm circumference (AC) using PCA-based forward stepwise regression (PCA-SWR) PCA-ANN, PCA-ANFIS and PCA-LS-SVM models were used to predict the effect of BP on post-unsupported Back reaction. METHODS: PCA was used to eliminate the multicollinearity between human predictor variables and the PC was selected from the original dataset. The selected main element is input to the established prediction model for modeling and testing. CONCLUSIONS: PCA-LS-SVM is more promising for predicting BP compared with other models by evaluating appropriate statistical indicators (deterministic coefficient, mean square root error, mean absolute percentage error). This assessment also demonstrates the importance and advancement of hybrid models in predicting other parameters in the biomedical field.