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.

On the use of restricted dissimilarity and dissimilarity-like functions for defining penalty functions

In this work we study the relation between restricted dissimilarity functions-and, more generally, dissimilarity-like functions- and penalty functions and the possibility of building the latter using the former. Several results on convexity and quasiconvexity are also considered.

Prevalence and stability of active play, restricted movement and television viewing in infants

This study describes engagement in and stability of physical activity and sedentary behaviours in early life, and assesses associations with sex, maternal education and developmental stage. Maternal-report data at child age 4, 9 and 20 months were collected from 542 families in the Melbourne Infant Feeding Activity and Nutrition Trial Program. Parents estimated average time per day their child spent in active pursuits or pursuits that restricted movement. With increasing age, children generally spent more time in active pursuits and watching television, and less time in situations that restrict movement. Associations were found with age of developmental milestone attainment but not sex or maternal education. Stability over time was strongest for television viewing (β = 0.34–0.38) and time spent outdoors (β = 0.27–0.33). Contrary to guidelines, television viewing increased and showed stability, suggesting a need to target this behaviour very early in life to achieve optimal longer term outcomes.

A systematic review of adherence to restricted diets in people with functional bowel disorders

Functional bowel disorders such as irritable bowel syndrome are commonly experienced within the population, and have an adverse impact on emotions, physical well-being, social activity, and occupational output. Adherence to a restricted diet can reduce symptoms, which in turn leads to increased quality of life and well-being. The aim of this review was to assess the extent to which predictors of dietary adherence have been considered in studies relating to functional bowel disorders and following a restricted diet. This was done firstly by examining such studies which contained a measure or indicator of adherence, and then by examining predictors of adherence within and between studies. A search of PsycINFO, Medline, CINAHL, Web of Science, and Cochrane databases was performed during July 2014, with the search criteria including relevant terms such as gastrointestinal disorder, irritable bowel syndrome, diet, and adherence. Of an initial 7927 papers, 39 were suitable for inclusion. Fourteen of the 39 studies included had a structured measure or indicator of dietary adherence, and the remaining 25 mentioned adherence without any structured levels of adherence. There was little investigation into the predictors of adherence, with symptom relief or induction being the primary goal of most of the studies. This review indicates that predictors of dietary adherence are rarely considered in research regarding functional bowel disorders. Further investigation is needed into the variables which contribute to rates of adherence to restricted diets, and more rigorous research is needed to characterise those individuals most likely to be non-adherent. Such research is necessary to ensure that people with these conditions can be provided with appropriate support and interventions.

Eating locally: Australasian gannets increase their foraging effort in a restricted range

During the breeding season, seabirds adopt a central place foraging strategy and are restricted in their foraging range by the fasting ability of their partner/chick and the cost of commuting between the prey resources and the nest. Because of the spatial and temporal variability of marine ecosystems, individuals must adapt their behaviour to increase foraging success within these constraints. The at-sea movements, foraging behaviour and effort of the Australasian gannet (Morus serrator) was determined over three sequential breeding seasons of apparent differing prey abundance to investigate how the species adapts to inter-annual fluctuations in food availability. GPS and tri-axial accelerometer data loggers were used to compare the degree of annual variation within two stages of breeding (incubation and chick rearing) at a small gannet colony situated between two larger, nearby colonies. Interestingly, neither males nor females increased the total distance travelled or duration of foraging trip in any breeding stage (P>0.05 in all cases) despite apparent low prey availability. However, consistently within each breeding stage, mean vectorial dynamic body acceleration (an index of energy expenditure) was greater in years of poorer breeding success (increased by a factor of three to eight), suggesting birds were working harder within their range. Additionally, both males and females increased the proportion of a foraging trip spent foraging in a poorer year across both breeding stages. Individuals from this colony may be limited in their ability to extend their range in years of low prey availability due to competition from conspecifics in nearby colonies and, consequently, increase foraging effort within this restricted foraging area.

Graph-induced restricted Boltzmann machines for document modeling

Discovering knowledge from unstructured texts is a central theme in data mining and machine learning. We focus on fast discovery of thematic structures from a corpus. Our approach is based on a versatile probabilistic formulation – the restricted Boltzmann machine (RBM) –where the underlying graphical model is an undirected bipartite graph. Inference is efficient document representation can be computed with a single matrix projection, making RBMs suitable for massive text corpora available today. Standard RBMs, however, operate on bag-of-words assumption, ignoring the inherent underlying relational structures among words. This results in less coherent word thematic grouping. We introduce graph-based regularization schemes that exploit the linguistic structures, which in turn can be constructed from either corpus statistics or domain knowledge. We demonstrate that the proposed technique improves the group coherence, facilitates visualization, provides means for estimation of intrinsic dimensionality, reduces overfitting, and possibly leads to better classification accuracy.

Tensor-variate restricted Boltzmann machines

Restricted Boltzmann Machines (RBMs) are an important class of latent variable models for representing vector data. An under-explored area is multimode data, where each data point is a matrix or a tensor. Standard RBMs applying to such data would require vectorizing matrices and tensors, thus resulting in unnecessarily high dimensionality and at the same time, destroying the inherent higher-order interaction structures. This paper introduces Tensor-variate Restricted Boltzmann Machines (TvRBMs) which generalize RBMs to capture the multiplicative interaction between data modes and the latent variables. TvRBMs are highly compact in that the number of free parameters grows only linear with the number of modes. We demonstrate the capacity of TvRBMs on three real-world applications: handwritten digit classification, face recognition and EEG-based alcoholic diagnosis. The learnt features of the model are more discriminative than the rivals, resulting in better classification performance.

Learning vector representation of medical objects via EMR-driven nonnegative restricted Boltzmann machines (eNRBM)

Electronic medical record (EMR) offers promises for novel analytics. However, manual feature engineering from EMR is labor intensive because EMR is complex - it contains temporal, mixed-type and multimodal data packed in irregular episodes. We present a computational framework to harness EMR with minimal human supervision via restricted Boltzmann machine (RBM). The framework derives a new representation of medical objects by embedding them in a low-dimensional vector space. This new representation facilitates algebraic and statistical manipulations such as projection onto 2D plane (thereby offering intuitive visualization), object grouping (hence enabling automated phenotyping), and risk stratification. To enhance model interpretability, we introduced two constraints into model parameters: (a) nonnegative coefficients, and (b) structural smoothness. These result in a novel model called eNRBM (EMR-driven nonnegative RBM). We demonstrate the capability of the eNRBM on a cohort of 7578 mental health patients under suicide risk assessment. The derived representation not only shows clinically meaningful feature grouping but also facilitates short-term risk stratification. The F-scores, 0.21 for moderate-risk and 0.36 for high-risk, are significantly higher than those obtained by clinicians and competitive with the results obtained by support vector machines.

Learning entry profiles of children with autism from multivariate treatment information using restricted boltzmann machines

Entry profiles can be generated before children with Autism Spectrum Disorders (ASD) begin to traverse an intervention program. They can help evaluate the progress of each child on the dedicated syllabus in addition to enabling narrowing down the best intervention course over time. However, the traits of ASD are expressed in different ways in every individual affected. The resulting spectrum nature of the disorder makes it challenging to discover profiles of children with ASD. Using data from 491 children, traversing the syllabus of a comprehensive intervention program on iPad called TOBY Playpad, we learn the entry profiles of the children based on their age, sex and performance on their first skills of the syllabus. Mixed-variate restricted Boltzmann machines allow us to integrate the heterogeneous data into one model making it a suitable technique. The data based discovery of entry profiles may assist in developing systems that can automatically suggest best suitable paths through the syllabus by clustering the children based on the characteristics they present at the beginning of the program. This may open the pathway for personalised intervention.

General equilibrium model with restricted participation in financial markets

The paper analyses a general equilibrium model with financiaI markets in which households may face restrictions in trading financiaI assets such as borrowing constraints and collateral (restricted participation model). However, markets are not assumed to be incomplete. We consider a standard general equilibrium model with H > 1 households, 2 periods and S states of nature in the second period. We show that generically the set of equilibrium allocations ia indeterminate, provided the existence of at least one nominal asset and one household for who some restriction is binding. Suppose there are C > 1 commodities in each state of nature and assets pays in units of some commodity. In this case for each household with binding restrictions it is possible to reduce the set of feasible assets trading and obtain a new equilibrium that utility improve alI those households. There is however an upper bound on the number of households to be improved related to the number of states of nature and the number of commodities. In particular, if the number of households ia smaller than the number of states of nature it is possible to Pareto improve any equilibrium by reducing the feasible choice set for each household.

Qualitative-feed-restricted heavy swine: meat quality and morpho-histochemical characteristics of muscle fibers

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Down-regulation of the cardiac sarcoplasmic reticulum ryanodine channel in severely food-restricted rats

We have shown that myocardial dysfunction induced by food restriction is related to calcium handling. Although cardiac function is depressed in food-restricted animals, there is limited information about the molecular mechanisms that lead to this abnormality. The present study evaluated the effects of food restriction on calcium cycling, focusing on sarcoplasmic Ca2+-ATPase (SERCA2), phospholamban (PLB), and ryanodine channel (RYR2) mRNA expressions in rat myocardium. Male Wistar-Kyoto rats, 60 days old, were submitted to ad libitum feeding (control rats) or 50% diet restriction for 90 days. The levels of left ventricle SERCA2, PLB, and RYR2 were measured using semi-quantitative RT-PCR. Body and ventricular weights were reduced in 50% food-restricted animals. RYR2 mRNA was significantly decreased in the left ventricle of the food-restricted group (control = 5.92 +/- 0.48 vs food-restricted group = 4.84 +/- 0.33, P < 0.01). The levels of SERCA2 and PLB mRNA were similar between groups (control = 8.38 +/- 0.44 vs food-restricted group = 7.96 +/- 0.45, and control = 1.52 +/- 0.06 vs food-restricted group = 1.53 +/- 0.10, respectively). Down-regulation of RYR2 mRNA expressions suggests that chronic food restriction promotes abnormalities in sarcoplasmic reticulum Ca2+ release.

Mechanical, biochemical, and morphological changes in the heart from chronic food-restricted rats

Food restriction (FR) has been shown to induce important morphological changes in rat myocardium. However, its influence on myocardial performance is not completely defined. We examined the effects of chronic FR on cardiac muscle function and morphology. Sixty-day-old Wistar-Kyoto rats were fed a control (C) or a restricted diet (daily intake reduced to 50% of the amount of food consumed by the control group) for 90 days. Myocardial performance was studied in isolated left ventricular (LV) papillary muscle. Fragments of the LV free wall were analysed by light microscopy, and the ultrastructure of the myocardium was examined in the LV papillary muscle. The myocardial collagen concentration was also evaluated. FR decreased body weight (BW) and LV weight (LVW); the LVW/BW ratio was higher in the restricted group (C, 1.86 +/- 0.17 mg/g; FR, 2.19 +/- 0.31 mg/g; p < 0.01). In the FR animals, the cardiac fibers were polymorphic, some of them were of small diameter and others presented lateral infoldings; the ultrastructural alterations were focal and included reduction of sarcoplasmic content, absence and (or) disorganization of myofilaments and Z line, numerous electron dense and polymorphic mitochondria, and deep infoldings of the plasma membrane. The hydroxyproline concentration was higher in the FR animals (p < 0.01). FR prolonged the contraction and relaxation time of the papillary muscle and did not change its ability to contract and shorten. In conclusion, although a 90-day period of FR caused striking myocardial ultrastructural alterations and increased the collagen concentration, it only minimally affected the mechanical function.