發(fā)布時(shí)間：2018-07-31 12:05

【摘要】：睡眠分期是醫(yī)學(xué)、神經(jīng)信息領(lǐng)域的研究熱點(diǎn)。人工標(biāo)記睡眠數(shù)據(jù)是一項(xiàng)費(fèi)時(shí)且費(fèi)力的工作。自動(dòng)睡眠分期方法能夠減少人工分期的工作負(fù)荷,但在復(fù)雜多變的臨床數(shù)據(jù)的應(yīng)用上仍存在局限性。本文提出了一種改進(jìn)的K均值聚類算法,主要目的是從實(shí)際睡眠數(shù)據(jù)的特點(diǎn)出發(fā),研究睡眠自動(dòng)分期方法。針對(duì)原始K均值聚類算法對(duì)初始聚類中心和離群點(diǎn)敏感的問題,本文結(jié)合密度的思想,選擇周圍數(shù)據(jù)密集的點(diǎn)作為初始中心,并根據(jù)"3σ法則"更新中心。改進(jìn)算法在健康被試和接受持續(xù)正壓通氣(CPAP)治療的睡眠障礙者的睡眠數(shù)據(jù)上進(jìn)行了測(cè)試,平均分類精確度達(dá)到76%,同時(shí)結(jié)合實(shí)際睡眠數(shù)據(jù)的形態(tài)多樣性驗(yàn)證討論了該方法在臨床數(shù)據(jù)上的可行性和有效性。
[Abstract]:Sleep staging is a hot topic in the field of medicine and neuroinformation. Manually marking sleep data is a time-consuming and laborious task. Automatic sleep staging method can reduce the workload of artificial staging, but there are still limitations in the application of complex and changeable clinical data. An improved K-means clustering algorithm is proposed in this paper. The main purpose of this algorithm is to study the automatic sleep staging method based on the characteristics of actual sleep data. In order to solve the problem that the original K-means clustering algorithm is sensitive to the initial clustering center and the outlier point, this paper combines the idea of density, selects the data dense points around as the initial center, and updates the center according to the "3 蟽 rule". The improved algorithm was tested on sleep data in healthy subjects and patients with sleep disorders treated with continuous positive pressure ventilation (CPAP). The average classification accuracy is 76 and the feasibility and effectiveness of the method in clinical data are discussed in combination with the morphological diversity of actual sleep data.
【作者單位】： 華東理工大學(xué)信息科學(xué)與工程學(xué)院自動(dòng)化系;上海諾城電氣有限公司;
【基金】：上海市自然科學(xué)基金項(xiàng)目資助(16ZR1407500) 上海市科委科技創(chuàng)新行動(dòng)計(jì)劃資助(12DZ1940903)
【分類號(hào)】：R740
，

本文編號(hào)：2155535