摘要
目的:为实现基于脉搏波的吸毒人员的快速甄别,对脉搏波甄别模型的构建进行研究。方法:应用采集、筛选的102例吸毒人员样本和138例非吸毒人员样本构建样本库,对从脉搏波中提取的数据进行分析,选择年龄、性别及slopeU、PWTT、AWX、WaveWidth6项作为模型特征,分别建立逻辑回归模型、支持向量机模型、随机森林模型。结果:逻辑回归模型、支持向量机模型、随机森林模型都可以较好地实现吸毒人员的甄别,其中随机森林模型性能相对更好。结论:通过采集脉搏波信息,运用数据挖掘技术,可以实现吸毒人员的甄别,今后可通过扩大样本容量、完善甄别模型,提高其应用可靠性。
Objective: To study on the construction of pulse wave - based screening model in order to realize the rapid screening of drug users. Methods: A total number of 102 drug users and 138 non - drug users were collected and screened as samples and the data extracted from the pulse wave were analyzed. Age,gender,slopeU,PWTT,AWX and WaveWidth were used as model features to construct respectively a Logistic regression model,a support vector machine model and a random forest model. Results: The Logistic regression model,support vector machine model and random forest model could better screen drug users,of which the performance of random forest model was the best. Conclusion: Drug users can be screened by pulse wave information and data mining technology. Further researches should be conducted to perfect the screening model and improve its reliability through collecting larger samples.
作者
顾海艳
王权
袁明
GU Haiyan;WANG Quan;YUAN Ming(Jiangsu Police Institute,Nanjing 210031,China;Shenzhen Qianhai National Health Technology Company,Shenzhen 51800,China)
出处
《东南大学学报（医学版）》
CAS
2019年第3期485-489,共5页
Journal of Southeast University(Medical Science Edition)
基金
江苏省“十三五”高等学校重点学科建设专项资金资助项目(20160838)
江苏省公安厅重点科研项目(2017KX033Z)
关键词
吸毒人员
脉搏波
甄别模型
逻辑回归
支持向量机
随机森林
drug users
pulse wave
screening model
logistic regression
support vector machine
random forest
