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.

Thurstonian Boltzmann machines: learning from multiple inequalities

We introduce Thurstonian Boltzmann Machines (TBM), a unified architecture that can naturally incorporate a wide range of data inputs at the same time. Our motivation rests in the Thurstonian view that many discrete data types can be considered as being generated from a subset of underlying latent continuous variables, and in the observation that each realisation of a discrete type imposes certain inequalities on those variables. Thus learning and inference in TBM reduce to making sense of a set of inequalities. Our proposed TBM naturally supports the following types: Gaussian, intervals, censored, binary, categorical, muticategorical, ordinal, (in)-complete rank with and without ties. We demonstrate the versatility and capacity of the proposed model on three applications of very different natures; namely handwritten digit recognition, collaborative filtering and complex social survey analysis. Copyright 2013 by the author(s).

A equação de Poisson-Boltzmann em regiões com fronteira irregular

Propomos uma idealização da situação em que uma macromolécula é ionizada em um solvente. Neste modelo a área da superfície da molécula é suposta ser grande com respeito a seu diâmetro. A molécula é considerada como um dielétrico com uma distribuição de cargas em sua superfície. Utilizando as condições de transmissão, a distribuição de Boltzmann no solvente e resultados recentes sobre espaços de Sobolev no contexto de espaços métricos, bem como de integração sobre superfícies irregulares, o problema é formulado em forma variacional. Resultados clássicos do cálculo de variações permitem a resolução analítica do problema.

Modelos cinéticos da equação linearizada de Boltzmann e um problema de transferência de calor em microescala

Neste trabalho, um problema de transferência de calor da dinâmica de gases rarefeitos, causado pela diferença de temperaturas nas superfícies de um canal, é abordado. O problema é formulado através dos modelos cinéticos BGK, S e Gross-Jackson da equação linearizada de Boltzmann e resolvido, de forma unificada, pelo método analítico de ordenadas discretas (método ADO). Resultados numéricos para as perturbações de densidade e temperatura e também para o fluxo de calor são apresentados e comparados, mostrando que não se pode dizer que algum dos três modelos seja uma melhor aproximação da solução aos resultados da equação linearizada de Boltzmann.

Protein-Ion Binding Process on Finite Macromolecular Concentration. A Poisson-Boltzmann and Monte Carlo Study

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

pK determination. A mean field, Poisson-Boltzmann approach

A model describing dissociation of monoprotonic acid and a method for the determination of its pK value are presented. The model is based on a mean field approximation. The Poisson-Boltzmann equation, adopting spherical symmetry, is numerically solved, and the solution of its linearized form is written. By use of the pH values of a dilution experiment of galacturonic acid as the entry data, the proposed method allowed estimation of the value of pK = 3.25 at a temperature of 25 degrees C. Values for the complex dimensions and dissociation degree are calculated using experimental pH values for solution concentration values ranging from 0.1 to 60 mM. The present analysis leads to the conclusion that the Poisson-Boltzmann equation or its linear form is equally suited for the description of such systems.

Podolsky electromagnetism at finite temperature: Implications on the Stefan-Boltzmann law

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Solution of the Poisson-Boltzmann equation for a system with four ionic species

The analytical solution of the Poisson-Boltzmann equation in an electrolyte with four ionic species (2:2:1:1), in the presence of a charged planar membrane or surface is presented. The function describing the mean electrical potential provides a convenient description that helps the understanding of electrical processes of biological interest.

INTEGRATION OF THE NONLINEAR POISSON BOLTZMANN-EQUATION FOR CHARGED VESICLES IN ELECTROLYTIC SOLUTIONS

The Poisson-Boltzmann equation (PBE), with specific ion-surface interactions and a cell model, was used to calculate the electrostatic properties of aqueous solutions containing vesicles of ionic amphiphiles. Vesicles are assumed to be water- and ion-permeable hollow spheres and specific ion adsorption at the surfaces was calculated using a Volmer isotherm. We solved the PBE numerically for a range of amphiphile and salt concentrations (up to 0.1 M) and calculated co-ion and counterion distributions in the inside and outside of vesicles as well as the fields and electrical potentials. The calculations yield results that are consistent with measured values for vesicles of synthetic amphiphiles.

pH at the micellar interface: Synthesis of pH probes derived from salicylic acid, acid-base dissociation in sodium dodecyl sulfate micelles, and Poisson-Boltzmann simulation

The study of the H+ concentration at the micellar interface is a convenient system for modeling the distribution of H+ at interfaces. We have synthesized salicylic acid derivatives to analyze the proton dissociation of both the carboxylic and phenol groups of' the probes, determining spectrophotometrically the apparent pK(a)'s (pK(ap)) in sodium dodecyl Sulfate, SDS, micelles with and without added salt. The synthesized probes were 2-hydroxy-5-(2-trimethylammoniumacetyl)benzoate; 2-hydroxy-5-(2-dimethylhexadecylammoniumacetyl)benzoate- 2-hydroxy-5-(2-dimethylhexadecylammoniumhexanoyl)benzoate-, 2-hydroxy-5-(2-diniethylhexadecylammoniumundecanoyl)betizoate; 2-hydroxy-5-acetylbenzoic acids and 2-hydroxy-5-dodecanoylbenzoic acid. Upon incorporation into SDS micelles the pK(ap)'s of both carboxylic and phenol groups increased by ca. 3 pH units and NaCl addition caused a decrease in the probe-incorporated pKap. The experimental results were fitted with a cell model Poisson-Boltzmann (P-B) equation taking in consideration the effect of salt on the aggregation number of SDS and using the distance of' the dissociating group as a parameter. The conformations of the probes were analyzed theoretically using two dielectric constants, e.g., 2 and 78. Both the P-B analysis and conformation calculations can be interpreted by assuming that the acid groups dissociate very close to, or at, the interface. Our results are consistent with the assumption that the intrinsic pK(a)'s of both carboxylic and phenol groups of the salicylic acid probes used here can be taken as those in water. Using this assumption the micellar and salt effects on the pKap's of the (trialkylammonium)benzoate probes were described accurately using a cell model P-B analysis. (c) 2005 Elsevier B.V. All rights reserved.

Diagnostic de securite d'une entreprise bresilienne a partir de l'analyse des accidents du travail par la methode de l'arbre des causes

We present here the results of a study of 21 work-related accidents that occurred in a Brazilian manufacturing company. The aim was to assess the safety level of the company to improve its work accident prevention policy. In the last 6 months of 1992 and 1993, all accidents resulting in 15 days' absence from work, reported for social security purposes, were analyzed using the INRS causal tree method (ADC) and a questionnaire completed on site. Potential risk factors for accidents were identified based on the specific factors highlighted by the ADC. More universal trees were also compiled for the safety assessment. Three hundred and thirty specific accident factors were recorded (man of 15.71 per accident). This is consistent with there being multiple causes of accidents rather than the assertion of Brazilian business safety departments that accidents are due to 'dangerous' or 'unsafe' behavior. Introducing the idea of culpability into accidents prevents the implementation of an appropriate information feedback process, essential for effective prevention. However, the large number of accidents related to 'material' (78%) and 'environment' (70%) indicates that working conditions are poor. This shows that the technical risks, mostly due to unsafe machinery and equipment are not being dealt with. Seventy-five potential accident factors were identified. Of these, 35% were 'organizational', a high proportion for the company studied. Improvisation occurs at all levels, particularly at the organizational level. This is, thus a major determinant for entire series of, if not most, accident situations. The poor condition of equipment also plays a major role in accidents. The effects of poor equipment on safety exacerbate the organizational shortcomings. The company's safety intervention policy should improve the management of human resources (rules designating particular workers for particular workstations; instructions for the safe operation of machines and equipment; training of operators, etc.) and introduce programs to detect risks and to improve the safety of machines and equipment. We also recommend the establishment of a program to follow the results of any preventive measures adopted.

Aplicação da equação de Paisson-Boltzmann com condição de contorno para regulação carga-potencial a sistemas polieletrolíticos

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)