Bayesian consensus clustering for multivariate longitudinal data

In clinical and epidemiological studies, there is a growing interest in studying the heterogeneity among patients based on longitudinal characteristics to identify subtypes of the study population. Compared to clustering a single longitudinal marker, simultaneously clustering multiple longitudinal markers allow additional information to be incorporated into the clustering process, which reveals co-existing longitudinal patterns and generates deeper biological insight. In the current study, we propose a Bayesian consensus clustering (BCC) model for multivariate longitudinal data. Instead of arriving at a single overall clustering, the proposed model allows each marker to follow marker-specific local clustering and these local clusterings are aggregated to find a global (consensus) clustering. To estimate the posterior distribution of model parameters, a Gibbs sampling algorithm is proposed. We apply our proposed model to the primary biliary cirrhosis study to identify patient subtypes that may be associated with their prognosis. We also perform simulation studies to compare the clustering performance between the proposed model and existing models under several scenarios. The results demonstrate that the proposed BCC model serves as a useful tool for clustering multivariate longitudinal data.     

Statistical analysis of gait patterns of persons with cerebral palsy

Patients at Boston's Children's Hospital diagnosed as having cerebral palsy were filmed walking. These films were digitized and translated into measurements associated with leg motion. In this paper we use the gait measurements of 128 such patients to illustrate that the kth nearest neighbour clustering procedure results in a gait typology for patients with cerebral palsy. The procedure identifies four subpopulations from the sample data; the membership of a patient within this typology is mostly determined by the patient's motor control. The developed typology differs from the present diagnostic system which classifies a cerebral palsy patient as either quadriplegic, diaplegic or hemiplegic.     

Clustering of autoimmune disease in parents of siblings from the Type 1 diabetes Warren repository.

AIMS: Autoimmune disorders co-exist in the same individuals and in families, implying a shared aetiology. The aim of this study was to compare the prevalence of the common autoimmune diseases in the parents of siblings from the Type 1 diabetes Warren repository with the general population. METHODS: Between 1989 and 1996, 505 British families with at least two siblings affected by Type 1 diabetes were recruited. Clinical information was collected regarding the presence of autoimmune disease in the parents and the prevalence of disease in the parents was compared with that expected in the general population. RESULTS: The prevalence of autoimmune disease in the parents was significantly higher in the repository compared with that expected in the general population [P-value = 1.98 x 10(-5) (female), P-value = 1.1 x 10(-8) (male)]. Type 1 diabetes was recorded in 63/1010 (6.2%) parents with a marked paternal preponderance (9.5 vs. 3%P = 0.002). Other autoimmune diseases affected 27% of parents with diabetes and 13.2% of parents without diabetes (P < 0.01). CONCLUSION: These data confirm the importance of family history as a significant risk factor for the development of Type 1 diabetes and support the hypothesis that the common autoimmune diseases share at least some aetiological mechanisms.     

基于实验均方误差F值的基因表达数据最佳聚类评判方法研究

目的 近年来产生了一些用于分析基因表达数据的聚类算法，却很少有关于评价聚类算法方法的研究。本研究的目的是尝试建立一个定量的评价基因表达数据聚类结果的方法。方法 本研究提供了一个系统的评价聚类结果的方法，利用我们提出的实验均方误差F值对几个常见的聚类算法进行比较。结果 利用F值对类质量的评价和利用已有的生物学知识对类进行分析的结果一致。结论 实验均方误差F值可以定量地评判用于基因表达数据的聚类算法。     

Short utterance recognition using a network with minimum training

A feedforward network is used to recognize short, digitized, isolated utterances. A high, multispeaker recognition rate is achieved with a small vocabulary with a single training utterance. This approach makes use of the pattern recognition property of the network architecture to classify different temporal patterns in the multidimensional feature space. The network recognizes the utterances without the need of segmentation, phoneme identification, or time alignment. We train the network with four words spoken by one single speaker. The network is then able to recognize 20 tokens spoken by 5 other speakers. We repeat the above training and testing procedure using a different speaker's utterances for training each time. The overall accuracy is 97.5%. We compare this approach to the traditional dynamic programming (DP) approach, and find that DP with slope constraints of 0 and 1 achieve 98.5% and 85% accuracies respectively. Finally we validate out statistics by training and testing the network of a four-word subset of the Texas Instruments (Tl) isolated word database. The accuracy with this vocabulary exceeds 96%. By doubling the size of the training set, the accuracy is raised to 98%. Using a suitable threshold, we are able to raise the accuracy of one network from 87% to 98.5%. Thresholding applied to all networks would then raise the overall accuracy to well over 99%.

This technique is especially promising because of the low overhead and computational requirements, which make it suitable for a low cost, portable, command recognition type of application.     

Unsupervised fuzzy clustering analysis supports behavioral cutoff criteria in an animal model of posttraumatic stress disorder.

BACKGROUND: Unsupervised fuzzy clustering (UFC) analysis is a mathematical technique that groups together objects in the multidimensional feature space according to a specified similarity measurement, thereby yielding clusters of similar data points that can be represented by a set of prototypes or centroids. METHODS: Since clinical studies of mental disorders distinguish between affected and unaffected individuals, we designed an inclusion/exclusion criteria (cutoff behavioral criteria [CBC]) approach for animal behavioral studies. The effect of classifying the study population into clearly affected versus clearly unaffected individuals according to behaviors on two behavioral paradigms was statistically significant. RESULTS: Here the raw data from previous studies were subjected to UFC algorithms as a means of objectively testing the validity of the concept of the CBC for our experimental model. The first UFC algorithm yielded two clearly discrete clusters, found to consist almost exclusively of the exposed animals in the one and unexposed animals in the other. The second algorithm yielded three clusters corresponding to animals designated as clearly affected, partially affected, and clearly unaffected. The algorithm for physiological data in addition to behavioral data failed to elicit discrete clusters. CONCLUSIONS: The UFC analysis yielded data that support the conceptual contention of the CBC and lends additional validity to our previous behavioral studies.     

Distribution of the largest neuron in mouse caudate-putamen nucleus: Its position in large-cell — Medium-cell clusters

Summary The position of the largest striatal neuron within territories delimited by medium-sized clustered neurons was charted in Nissl-stained sections through the mouse caudate-putamen nucleus. Medium-sized neuron somata occur in close proximity to this large cell at some point in the anteroposterior, mediolateral or dorsoventral extent of its soma. The size of the network of medium-sized neurons associated with the large cell may vary from two to 15 neurons. Even when this network is extensive, the large neuron is never completely surrounded. Most often, this cell also borders a fascicle of internal capsule fibers, and the entire cellular island may be aligned either parallel to or perpendicular to the orientation of these fibers. These findings suggest the hypothesis that cellular territories in the caudate-putamen nucleus have a very specific orientation in three dimensional space.     

Clustering of allergenic pollen on the basis of skin responses of atopic patients by matrix analysis

The responses of 148 atopic patients to some 43 different extracts of allergenic pollen were tested by prick tests. The measure of dissimilarity was introduced and calculated for all pairs of allergens. The investigated allergens were clustered into groups, according to their unbiased greatest similarity, by a matrix-structuring method. Results indicate that subgroups of allergens can be distinguished even within groups of closely related pollen allergens that were believed to be fully cross-reactive. A few cases are demonstrated for various varieties of olives, pecans, date palms, and turf grasses and for some wild chenopods and amaranths. The usefulness of the suggested solution for allergy research and for clinical practice is discussed.     

Spatiotemporal analysis of event-related potentials to upright, inverted, and contrast-reversed faces: effects on encoding and recognition

In an n-back face recognition task where subjects responded to repeated stimuli, ERPs were recorded to upright, inverted, and contrast-reversed faces. The effects of inversion and contrast reversal on face encoding and recognition were investigated using the multivariate spatiotemporal partial least squares (PLS) analysis. The configural manipulations affected early processing (100-200 ms) at posterior sites: Inversion effects were parietal and lateral, whereas contrast-reversal effects were more occipital and medial, suggesting different underlying generators. A later reactivation of face processing areas was unique to inverted faces, likely due to processing difficulties. PLS also indicated that the "old-new" repetition effect was maximal for upright faces and likely involved frontotemporal areas. Marked processing differences between inverted and contrast-reversed faces were seen, but these effects were similar at encoding and recognition.     

Geographic patterns and socioeconomic differences in the nutritional quality of household packaged food purchases in the United States

BackgroundIt is unclear whether dietary quality varies by geography in the US.PurposeAssess patterns in packaged food purchases (PFPs).MethodsWe characterized variation in PFP quality from 2008 to 2018 by 1) examining geographic clustering and 2) using regression analysis to control for household characteristics.ResultsLower quality purchases clustered in the Southeast and Appalachia, whereas higher quality purchases clustered in the West and Northeast. Spatial patterns were similar for low socioeconomic households but not high socioeconomic households. Geographic differences in quality remained after controlling for demographic composition.ConclusionThis analysis should inform research into systemic drivers of PFP quality.