2017286-1
专题:物联网技术与应用
基于 K-means 和 MTLS-SVM 算法的生理参数监测系统
夏景明,唐玲玲,谈玲,郑晗
(南京信息工程大学,江苏 南京 210044)
摘 要:在非医模式的生理参数监测系统中,对监测参数进行学习,可以提高诊断和预测精度。针对多任务
时间序列中存在的信息挖掘不充分、预测精度低等问题,将机器学习中的监督和半监督学习方式结合起来对
远程健康监护对象进行生理状况预测。该方法用 K-means 算法将相同类别的数据集群,并使用多任务最小二
乘支持向量机(MTLS-SVM)来训练历史数据来进行趋势预测。为了评估该方法的有效性,将 MTLS-SVM 方
法与 K-means、MTLS-SVM 方法比较,实验结果表明该方法具有较高的预测精度。
关键词:生理参数;时间序列预测;K-means 聚类;多任务学习
中图分类号:TN919 文献标识码:A
doi: 10.11959/j.issn.1000-0801.2017286
Biometric monitoring system based on K-means &
MTLS-SVM algorithm
XIA Jingming, TANG Lingling, TAN Ling, ZHENG Han
Nanjing University of Information Science & Technology, Nanjing 210044, China
Abstract: In a nonmedical biometric monitoring system, the monitoring parameters are preceded with machine
learning for precision promotion of diagnosis and prediction. Considering the problems of insufficient information
mining and low prediction accuracy in multi task time series, both supervised and unsupervised machine learning
techniques were applied to predict the physical condition of the remote health care. These techniques were K-means
for clustering the similar group of data and MTLS-SVM model for training and testing historical data to perform a
trend prediction. In order to evaluate the effectiveness of the method, the proposed method was compared with
MTLS-SVM method. The experimental results show that the proposed method has higher prediction accuracy.
Key words: physiological parameter, time series prediction, K-means clustering, multi-task learning
1 引言
我国社会开始进入老年化阶段,老龄化趋势
明显,患有慢性病的人群(如患有心脏病和血压
不稳定人群)也逐渐增长。老化性患者和慢性病
患者需要进行长时间的观察与医治,但由于条件
的限制及其他因素,这类人群通常是按期到医院
复查诊治,那么对病人的治疗效果会不佳,除人
收稿日期:2017-09-01;修回日期:2017-09-30
基金项目:国家自然科学基金资助项目(No.41505017);江苏省自然科学基金资助项目(No.BK20160951)
Foundation Items: The National Natural Science Foundation of China (No.41505017), The National Natural Science Foundation o
Jiangsu Province of China (No.BK20160951)