Clustering and variable selection in the presence of mixed variable types and missing data

We consider the problem of model‐based clustering in the presence of many correlated, mixed continuous, and discrete variables, some of which may have missing values. Discrete variables are treated with a latent continuous variable approach, and the Dirichlet process is used to construct a mixture model with an unknown number of components. Variable selection is also performed to identify the variables that are most influential for determining cluster membership. The work is motivated by the need to cluster patients thought to potentially have autism spectrum disorder on the basis of many cognitive and/or behavioral test scores. There are a modest number of patients (486) in the data set along with many (55) test score variables (many of which are discrete valued and/or missing). The goal of the work is to (1) cluster these patients into similar groups to help identify those with similar clinical presentation and (2) identify a sparse subset of tests that inform the clusters in order to eliminate unnecessary testing. The proposed approach compares very favorably with other methods via simulation of problems of this type. The results of the autism spectrum disorder analysis suggested 3 clusters to be most likely, while only 4 test scores had high (>0.5) posterior probability of being informative. This will result in much more efficient and informative testing. The need to cluster observations on the basis of many correlated, continuous/discrete variables with missing values is a common problem in the health sciences as well as in many other disciplines.     

Spectral synchronicity in brain signals

This paper addresses the problem of identifying brain regions with similar oscillatory patterns detected from electroencephalograms. We introduce the hierarchical spectral merger (HSM) clustering method where the feature of interest is the spectral curve and the similarity metric used is the total variance distance. The HSM method is compared with clustering using features derived from independent‐component analysis. Moreover, the HSM method is applied to 2 different electroencephalogram datasets. The first was recorded at resting state where the participant was not engaged in any cognitive task; the second was recorded during a spontaneous epileptic seizure. The results of the analyses using the HSM method demonstrate that clustering could evolve over the duration of the resting state and during epileptic seizure.     

Multivariate generalized hidden Markov regression models with random covariates: Physical exercise in an elderly population

A time‐varying latent variable model is proposed to jointly analyze multivariate mixed‐support longitudinal data. The proposal can be viewed as an extension of hidden Markov regression models with fixed covariates (HMRMFCs), which is the state of the art for modelling longitudinal data, with a special focus on the underlying clustering structure. HMRMFCs are inadequate for applications in which a clustering structure can be identified in the distribution of the covariates, as the clustering is independent from the covariates distribution. Here, hidden Markov regression models with random covariates are introduced by explicitly specifying state‐specific distributions for the covariates, with the aim of improving the recovering of the clusters in the data with respect to a fixed covariates paradigm. The hidden Markov regression models with random covariates class is defined focusing on the exponential family, in a generalized linear model framework. Model identifiability conditions are sketched, an expectation‐maximization algorithm is outlined for parameter estimation, and various implementation and operational issues are discussed. Properties of the estimators of the regression coefficients, as well as of the hidden path parameters, are evaluated through simulation experiments and compared with those of HMRMFCs. The method is applied to physical activity data.     

Clustering patterns of type-type combination in multiple genotypes infections of human papillomavirus in cervical adenocarcinoma

Coinfections with multiple types of human papillomavirus (HPV) occur in cervical adenocarcinoma (ADC). However, it remains unclear the clustering patterns of multiple types in HPV coinfections and relevant factors in ADC. A total of 718 paraffin-embedded ADC specimens were collected in China and tested for HPV genotypes using SPF10-INNO-LIPA. The prevalence of HPV coinfections and clustering patterns by geographical regions, histological subtypes and ages were assessed. Type-specific attribution of HPV to ADC adjusted by HPV coinfections were calculated. The prevalence of HPV coinfections was found to be 8.4% in ADC cases with slight variation by geographic regions between 2.2% and 12.5%. The 88.3% of all HPV multiple infections in ADC were two types of HPV coinfections with predominant combination of HPV 16 and HPV 18. The attribution to ADC was 88.0% for HPV 16/18 targeted by bivalent and quadrivalent vaccine and 96.8% for HPV 16/18/31/33/45/52/58 targeted by nonavalent vaccine. Clustering patterns of multiple types were related with histological categories and age at diagnosis. In conclusion, HPV coinfections are uncommonly prevalent in ADC cases with slight variation by geographic regions and distinct clustering patterns of multiple types by histological subtypes and ages at diagnosis. The high attribution of carcinogenic HPV types to ADC predicts potential protection of HPV vaccine against ADC.     

基于直方图均衡化的伽马校正和K-means聚类的舌象苔质分离方法

目的舌苔、舌质分离对后续肿瘤患者舌象的客观化辨证具有重要的意义。常用的算法是基于颜色空间通道舌图像的K-means聚类算法。CIELAB颜色空间的a*通道舌图像相较于其他颜色空间通道的舌图像分割结果稳定,常用于后续的分割。对于部分舌图像,a*通道舌图像的舌苔、舌质虽然具有一定的区分度,但二者的区分度并不是十分明显,影响后续的分割结果。因此,本文提出一种基于直方图均衡化的伽马校正和K-means聚类的舌苔、舌质分离方法。方法采用200幅肿瘤患者的舌图像作为实验样本。首先将舌图像从RGB颜色空间转换到CIELAB颜色空间,对a*通道舌图像进行直方图均衡化增强以及伽马校正,然后利用K-means聚类方法对增强后的舌图像舌苔、舌质分离,得到直方图均衡化以及伽马校正后的a*通道舌图像和分割后的舌苔、舌质图像。为了验证算法的可行性,请5位专业中医医生对200例肿瘤患者的舌图像舌苔、舌质分割效果进行辨析。结果进行直方图均衡化以及伽马校正后的a*通道舌图像舌苔、舌质分割结果明显强于未经处理的a*通道舌图像分割结果。经辨析,分割合格率为97%。结论该方法可以很好地实现舌苔、舌质分离,具有一定的应用价值。     

Psychosocial Risk Factors for Hospitalisation and Death from Chronic Obstructive Pulmonary Disease: A Prospective Cohort Study

Only a few smaller studies have addressed the effect of psychosocial factors on risk of chronic obstructive pulmonary disease (COPD) in spite of the potential for psychosocial stress to affect development of the disease through immunological and behavioural pathways. The aim of this study is to determine the relation between various psychosocial risk factors, individually and accumulated, and COPD hospitalisation and deaths. A total of 8728 women and men free of asthma and COPD participating in the Copenhagen City Heart Study, were asked comprehensive questions on major life events, work-related stress, social network, vital exhaustion, economic hardship, and sleep medication in 1991–1993 and followed in nationwide registers until 2009, with <2% loss to follow-up. During follow-up, 461 women and 352 men were hospitalized with or died from COPD. Major life events in adult life and vital exhaustion were both associated with a higher risk of COPD in an exposure-dependent manner, with high vital exhaustion being associated with a hazard ratio [HR] of 2.31 (95% CI 1.69–3.16) for women and 2.48 (1.69–3.64) for men. A higher risk of COPD was also found in participants who experienced economic hardship or had a dysfunctional social network. Furthermore, the accumulation of psychosocial risk factors was associated with a higher risk of COPD in both women (HR = 2.40, 1.78–3.22) and men (HR = 1.93, 1.33–2.80). Psychosocial vulnerability may be important to consider both in clinical practice and when planning future preventive strategies against COPD.     

全国各省3种主要性传播疾病发病率的聚类分析

目的利用层次聚类分析法,对我国各省区3种主要性传播疾病发病率进行综合评判分类。方法 借助SPSS 15.0对我国2008年31个省区市3种主要性病-艾滋病、淋病、梅毒的发病率资料进行Q型聚类分析,距离测量采用组间平均法。结果当各省区市性病发病情况聚为4类时,广西,上海和浙江,云南和河南,因其不同的性病高发特点,各聚为一类,其余省区市聚为一类,未出现单个或少数几个省区因疾病低发而单独聚为一类的情况。结论我国性病的发病存在明显的高发区,而低发区发病特点不明显,更接近一般发病水平。