Classification of main meal patterns - a latent class approach

Relatively little examination of the meals that are prepared in households has been conducted, despite their well-defined properties and widespread community interest in their preparation. The purpose of the present study was to identify the patterns of main meal preparation among Australian adult household meal preparers aged 44 years and younger and 45 years and over, and the relationships between these patterns and likely socio-demographic and psychological predictors. An online cross-sectional survey was conducted by Meat and Livestock Australia among a representative sample of people aged 18–65 years in Australia in 2011. A total of 1076 usable questionnaires were obtained, which included categorical information about the main meal dishes that participants had prepared during the previous 6 months along with demographic information, the presence or absence of children at home, confidence in seasonal food knowledge and personal values. Latent class analysis was applied and four types of usage patterns of thirty-three popular dishes were identified for both age groups, namely, high variety, moderate variety, high protein but low beef and low variety. The meal patterns were associated differentially with the covariates between the age groups. For example, younger women were more likely to prepare a high or moderate variety of meals than younger men, while younger people who had higher levels of education were more likely to prepare high-protein but low-beef meals. Moreover, young respondents with higher BMI were less likely to prepare meals with high protein but low beef content. Among the older age group, married people were more likely to prepare a high or moderate variety of meals than people without partners. Older people who held strong universalist values were more likely to prepare a wide variety of meals with high protein but low beef content. For both age groups, people who had children living at home and those with better seasonal food knowledge were more likely to prepare a high variety of dishes. The identification of classes of meal users would enable health communication to be tailored to improve meal patterns. Moreover, the concept of meals may be useful for health promotion, because people may find it easier to change their consumption of meals rather than individual foods.

Latent patient profile modelling and applications with mixed-variate restricted Boltzmann machine

Efficient management of chronic diseases is critical in modern health care. We consider diabetes mellitus, and our ongoing goal is to examine how machine learning can deliver information for clinical efficiency. The challenge is to aggregate highly heterogeneous sources including demographics, diagnoses, pathologies and treatments, and extract similar groups so that care plans can be designed. To this end, we extend our recent model, the mixed-variate restricted Boltzmann machine (MV.RBM), as it seamlessly integrates multiple data types for each patient aggregated over time and outputs a homogeneous representation called "latent profile" that can be used for patient clustering, visualisation, disease correlation analysis and prediction. We demonstrate that the method outperforms all baselines on these tasks - the primary characteristics of patients in the same groups are able to be identified and the good result can be achieved for the diagnosis codes prediction.

Learning sparse latent representation and distance metric for image retrieval

The performance of image retrieval depends critically on the semantic representation and the distance function used to estimate the similarity of two images. A good representation should integrate multiple visual and textual (e.g., tag) features and offer a step closer to the true semantics of interest (e.g., concepts). As the distance function operates on the representation, they are interdependent, and thus should be addressed at the same time. We propose a probabilistic solution to learn both the representation from multiple feature types and modalities and the distance metric from data. The learning is regularised so that the learned representation and information-theoretic metric will (i) preserve the regularities of the visual/textual spaces, (ii) enhance structured sparsity, (iii) encourage small intra-concept distances, and (iv) keep inter-concept images separated. We demonstrate the capacity of our method on the NUS-WIDE data. For the well-studied 13 animal subset, our method outperforms state-of-the-art rivals. On the subset of single-concept images, we gain 79:5% improvement over the standard nearest neighbours approach on the MAP score, and 45.7% on the NDCG.

Extraction of latent patterns and contexts from social honest signals using hierarchical Dirichlet processes

A fundamental task in pervasive computing is reliable acquisition of contexts from sensor data. This is crucial to the operation of smart pervasive systems and services so that they might behave efficiently and appropriately upon a given context. Simple forms of context can often be extracted directly from raw data. Equally important, or more, is the hidden context and pattern buried inside the data, which is more challenging to discover. Most of existing approaches borrow methods and techniques from machine learning, dominantly employ parametric unsupervised learning and clustering techniques. Being parametric, a severe drawback of these methods is the requirement to specify the number of latent patterns in advance. In this paper, we explore the use of Bayesian nonparametric methods, a recent data modelling framework in machine learning, to infer latent patterns from sensor data acquired in a pervasive setting. Under this formalism, nonparametric prior distributions are used for data generative process, and thus, they allow the number of latent patterns to be learned automatically and grow with the data - as more data comes in, the model complexity can grow to explain new and unseen patterns. In particular, we make use of the hierarchical Dirichlet processes (HDP) to infer atomic activities and interaction patterns from honest signals collected from sociometric badges. We show how data from these sensors can be represented and learned with HDP. We illustrate insights into atomic patterns learned by the model and use them to achieve high-performance clustering. We also demonstrate the framework on the popular Reality Mining dataset, illustrating the ability of the model to automatically infer typical social groups in this dataset. Finally, our framework is generic and applicable to a much wider range of problems in pervasive computing where one needs to infer high-level, latent patterns and contexts from sensor data.

The value of impulsivity to define subgroups of addicted individuals differing in personality dysfunction, craving, psychosocial adjustment, and wellbeing: A latent class analysis

High impulsivity is common to substance and gambling addictions. Despite these commonalities, there is still substantial heterogeneity on impulsivity levels within these diagnostic groups, and variations in impulsive levels predict higher severity of symptoms and poorer outcomes. We addressed the question of whether impulsivity scores can yield empirically driven subgroups of addicted individuals that will exhibit different clinical presentations and outcomes. We applied latent class analysis (LCA) to trait (UPPS-P impulsive behavior scale) and cognitive impulsivity (Stroop and d2 tests) scores in three predominantly male addiction diagnostic groups: Cocaine with Personality Disorders, Cocaine Non-comorbid, and Gambling and analyzed the usefulness of the resulting subgroups to differentiate personality beliefs and relevant outcomes: Craving, psychosocial adjustment, and quality of life. In accordance with impulsivity scores, the three addiction diagnostic groups are best represented as two separate classes: Class 1 characterized by greater trait impulsivity and poorer cognitive impulsivity performance and Class 2 characterized by lower trait impulsivity and better cognitive impulsivity performance. The two empirically derived classes showed significant differences on personality features and outcome variables (Class 1 exhibited greater personality dysfunction and worse clinical outcomes), whereas conventional diagnostic groups showed non-significant differences on most of these measures. Trait and cognitive impulsivity scores differentiate subgroups of addicted individuals with more versus less severe personality features and clinical outcomes.

Learning latent activities from social signals with hierarchical dirichlet processes

Understanding human activities is an important research topic, most noticeably in assisted-living and healthcare monitoring environments. Beyond simple forms of activity (e.g., an RFID event of entering a building), learning latent activities that are more semantically interpretable, such as sitting at a desk, meeting with people, or gathering with friends, remains a challenging problem. Supervised learning has been the typical modeling choice in the past. However, this requires labeled training data, is unable to predict never-seen-before activity, and fails to adapt to the continuing growth of data over time. In this chapter, we explore the use of a Bayesian nonparametric method, in particular the hierarchical Dirichlet process, to infer latent activities from sensor data acquired in a pervasive setting. Our framework is unsupervised, requires no labeled data, and is able to discover new activities as data grows. We present experiments on extracting movement and interaction activities from sociometric badge signals and show how to use them for detecting of subcommunities. Using the popular Reality Mining dataset, we further demonstrate the extraction of colocation activities and use them to automatically infer the structure of social subgroups. © 2014 Elsevier Inc. All rights reserved.

Measurement of stable changes of self-management skills after rehabilitation: a latent state-trait analysis of the Health Education Impact Questionnaire (heiQ™)

To assess stable effects of self-management programs, measurement instruments should primarily capture the attributes of interest, for example, the self-management skills of the measured persons. However, measurements of psychological constructs are always influenced by both aspects of the situation (states) and aspects of the person (traits). This study tests whether the Health Education Impact Questionnaire (heiQ™), an instrument assessing a wide range of proximal outcomes of self-management programs, is primarily influenced by person factors instead of situational factors. Furthermore, measurement invariance over time, changes in traits and predictors of change for each heiQ™ scale were examined.

Measurement of stable changes of self-management skills after rehabilitation: a latent state–trait analysis of the Health Education Impact Questionnaire (heiQ™)

Methods: Subjects were N = 580 patients with rheumatism, asthma, orthopedic conditions or inflammatory bowel disease, who filled out the heiQ™ at the beginning, the end of and 3 months after a disease-specific inpatient rehabilitation program in Germany. Structural equation modeling techniques were used to estimate latent trait-change models and test for measurement invariance in each heiQ™ scale. Coefficients of consistency, occasion specificity and reliability were computed.

Multichannel biomedical time series clustering via hierarchical probabilistic latent semantic analysis

Biomedical time series clustering that automatically groups a collection of time series according to their internal similarity is of importance for medical record management and inspection such as bio-signals archiving and retrieval. In this paper, a novel framework that automatically groups a set of unlabelled multichannel biomedical time series according to their internal structural similarity is proposed. Specifically, we treat a multichannel biomedical time series as a document and extract local segments from the time series as words. We extend a topic model, i.e., the Hierarchical probabilistic Latent Semantic Analysis (H-pLSA), which was originally developed for visual motion analysis to cluster a set of unlabelled multichannel time series. The H-pLSA models each channel of the multichannel time series using a local pLSA in the first layer. The topics learned in the local pLSA are then fed to a global pLSA in the second layer to discover the categories of multichannel time series. Experiments on a dataset extracted from multichannel Electrocardiography (ECG) signals demonstrate that the proposed method performs better than previous state-of-the-art approaches and is relatively robust to the variations of parameters including length of local segments and dictionary size. Although the experimental evaluation used the multichannel ECG signals in a biometric scenario, the proposed algorithm is a universal framework for multichannel biomedical time series clustering according to their structural similarity, which has many applications in biomedical time series management.

On the strength of dislocation interactions and their effect on latent hardening in pure Magnesium

This study is dedicated to the quantification of latent hardening and its effect on the plasticity of pure hexagonal magnesium. To this end, discrete dislocation dynamics simulations are used to (1) extract latent hardening parameters coupling different slip systems, and to (2) assess the validity of two existing constitutive models linking slip system strength to dislocation densities on all slip systems. As hexagonal materials deform via activation of different slip modes, each with different mobilities and lattice friction stress, the effects of the latter on latent hardening evolution are also investigated. It is found that the multi-slip formulation proposed by Franciosi and Zaoui gives accurate predictions when multiple interactions are involved while the formulation suggested by Lavrentev and Pokhil systematically overestimates the flow stress. Similar to FCC materials, it is also found that collinear interactions potentially contribute the most to latent hardening. Basal/pyramidal c + a interactions are found to be very strong, while interactions involving second-order pyramidal c + a primary dislocations appear to be the weakest ones. Finally, the latent hardening parameters, extracted from the discrete dislocation dynamics simulations, are used in polycrystal simulations and the impact of finely accounting for latent hardening on predictions of the macroscopic anisotropic response is shown to be of significant importance.