Classification and staging of chronic liver disease from multimodal data

Chronic liver disease (CLD) is most of the time an asymptomatic, progressive, and ultimately potentially fatal disease. In this study, an automatic hierarchical procedure to stage CLD using ultrasound images, laboratory tests, and clinical records are described. The first stage of the proposed method, called clinical based classifier (CBC), discriminates healthy from pathologic conditions. When nonhealthy conditions are detected, the method refines the results in three exclusive pathologies in a hierarchical basis: 1) chronic hepatitis; 2) compensated cirrhosis; and 3) decompensated cirrhosis. The features used as well as the classifiers (Bayes, Parzen, support vector machine, and k-nearest neighbor) are optimally selected for each stage. A large multimodal feature database was specifically built for this study containing 30 chronic hepatitis cases, 34 compensated cirrhosis cases, and 36 decompensated cirrhosis cases, all validated after histopathologic analysis by liver biopsy. The CBC classification scheme outperformed the nonhierachical one against all scheme, achieving an overall accuracy of 98.67% for the normal detector, 87.45% for the chronic hepatitis detector, and 95.71% for the cirrhosis detector.

Categorical data clustering using a minimum message length criterion

Research on cluster analysis for categorical data continues to develop, new clustering algorithms being proposed. However, in this context, the determination of the number of clusters is rarely addressed. We propose a new approach in which clustering and the estimation of the number of clusters is done simultaneously for categorical data. We assume that the data originate from a finite mixture of multinomial distributions and use a minimum message length criterion (MML) to select the number of clusters (Wallace and Bolton, 1986). For this purpose, we implement an EM-type algorithm (Silvestre et al., 2008) based on the (Figueiredo and Jain, 2002) approach. The novelty of the approach rests on the integration of the model estimation and selection of the number of clusters in a single algorithm, rather than selecting this number based on a set of pre-estimated candidate models. The performance of our approach is compared with the use of Bayesian Information Criterion (BIC) (Schwarz, 1978) and Integrated Completed Likelihood (ICL) (Biernacki et al., 2000) using synthetic data. The obtained results illustrate the capacity of the proposed algorithm to attain the true number of cluster while outperforming BIC and ICL since it is faster, which is especially relevant when dealing with large data sets.

The influence of inter-domain mobility on message stream response time in wired/wireless PROFIBUS-based networks

In previous works we have proposed a hybrid wired/wireless PROFIBUS solution where the interconnection between the heterogeneous media was accomplished through bridge-like devices with wireless stations being able to move between different wireless cells. Additionally, we had also proposed a worst-case timing analysis assuming that stations were stationary. In this paper we advance these previous works by proposing a worst-case timing analysis for the system’s message streams considering the effect of inter-cell mobility.

Constructing and Using Multimodal Narratives to Research in Science Education: Contributions Based on Practical Classroom

This study deals with the problem of how to collect genuine and useful data about science classroom practices, and preserving the complex and holistic nature of teaching and learning. Additionally, we were looking for an instrument that would allow comparability and verifiability for teaching and research purposes. Given the multimodality of teaching and learning processes, we developed the multimodal narrative (MN), which describes what happens during a task and incorporates data such as examples of students’ work.

Evaluating P-NET message's response time with fixed priority queuing at application process level

P-NET is a multi-master fieldbus standard based on a virtual token passing scheme. In P-NET each master is allowed to transmit only one message per token visit. In the worst-case, the communication response time can be derived considering that, in each token cycle, all stations use the token to transmit a message. In this paper, we define a more sophisticated P-NET model, which considers the actual token utilisation. We then analyse the possibility of implementing a local priority-based scheduling policy to improve the real-time behaviour of P-NET.

From task scheduling in single processor environments to message scheduling in a PROFIBUS fieldbus network

In this paper we survey the most relevant results for the prioritybased schedulability analysis of real-time tasks, both for the fixed and dynamic priority assignment schemes. We give emphasis to the worst-case response time analysis in non-preemptive contexts, which is fundamental for the communication schedulability analysis. We define an architecture to support priority-based scheduling of messages at the application process level of a specific fieldbus communication network, the PROFIBUS. The proposed architecture improves the worst-case messages’ response time, overcoming the limitation of the first-come-first-served (FCFS) PROFIBUS queue implementations.

Evaluating the duration of message transactions in broadcast wired/wireless fieldbus networks

Determining the response time of message transactions is one of the major concerns in the design of any distributed computer-controlled system. Such response time is mainly dependent on the medium access delay, the message length and the transmission delay. While the medium access delay in fieldbus networks has been thoroughly studied in the last few years, the transmission delay has been almost ignored as it is considered that it can be neglected when compared to the length of the message itself. Nevertheless, this assumption is no longer valid when considering the case of hybrid wired/wireless fieldbus networks, where the transmission delay through a series of different mediums can be several orders of magnitude longer than the length of the message itself. In this paper, we show how to compute the duration of message transactions in hybrid wired/wireless fieldbus networks. This duration is mainly dependent on the duration of the request and response frames and on the number and type of physical mediums that the frames must cross between initiator and responder. A case study of a hybrid wired/wireless fieldbus network is also presented, where it becomes clear the interest of the proposed approach

Guaranteeing real-time message deadlines in PROFIBUS networks

The paper provides a comprehensive study on how to use Profibus networks to support real time communications, that is, ensuring the transmission of the real time messages before their deadlines. Profibus is based on a simplified Timed Token (TT) protocol, which is a well proved solution for real time communication systems. However, Profibus differences from the TT protocol prevent the application of the usual TT analysis. The main reason is that, conversely to the TT protocol, in the worst case, only one high priority message is processed per token visit. The major contribution of the paper is to prove that, despite this shortcoming, it is possible to guarantee communication real time behaviour with the Profibus protocol

Enhancing Multimodal Silent Speech Interfaces With Feature Selection

In research on Silent Speech Interfaces (SSI), different sources of information (modalities) have been combined, aiming at obtaining better performance than the individual modalities. However, when combining these modalities, the dimensionality of the feature space rapidly increases, yielding the well-known "curse of dimensionality". As a consequence, in order to extract useful information from this data, one has to resort to feature selection (FS) techniques to lower the dimensionality of the learning space. In this paper, we assess the impact of FS techniques for silent speech data, in a dataset with 4 non-invasive and promising modalities, namely: video, depth, ultrasonic Doppler sensing, and surface electromyography. We consider two supervised (mutual information and Fisher's ratio) and two unsupervised (meanmedian and arithmetic mean geometric mean) FS filters. The evaluation was made by assessing the classification accuracy (word recognition error) of three well-known classifiers (knearest neighbors, support vector machines, and dynamic time warping). The key results of this study show that both unsupervised and supervised FS techniques improve on the classification accuracy on both individual and combined modalities. For instance, on the video component, we attain relative performance gains of 36.2% in error rates. FS is also useful as pre-processing for feature fusion. Copyright © 2014 ISCA.

Sistema multimodal para captura e anotação de vídeo

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Message-oriented middleware for smart grids

In order to increase the efficiency in the use of energy resources, the electrical grid is slowly evolving into a smart(er) grid that allows users' production and storage of energy, automatic and remote control of appliances, energy exchange between users, and in general optimizations over how the energy is managed and consumed. One of the main innovations of the smart grid is its organization over an energy plane that involves the actual exchange of energy, and a data plane that regards the Information and Communication Technology (ICT) infrastructure used for the management of the grid's data. In the particular case of the data plane, the exchange of large quantities of data can be facilitated by a middleware based on a messaging bus. Existing messaging buses follow different data management paradigms (e.g.: request/response, publish/subscribe, data-oriented messaging) and thus satisfy smart grids' communication requirements at different extents. This work contributes to the state of the art by identifying, in existing standards and architectures, common requirements that impact in the messaging system of a data plane for the smart grid. The paper analyzes existing messaging bus paradigms that can be used as a basis for the ICT infrastructure of a smart grid and discusses how these can satisfy smart grids' requirements.

Implementation of an IoT-based Sensor Network Integrating IoT-based Sensor Networks with Large-Scale Message Distribution

As technology advances not only do new standards and programming styles appear but also some of the previously established ones gain relevance. In a new Internet paradigm where interconnection between small devices is key to the development of new businesses and scientific advancement there is the need to find simple solutions that anyone can implement in order to allow ideas to become more than that, ideas. Open-source software is still alive and well, especially in the area of the Internet of Things. This opens windows for many low capital entrepreneurs to experiment with their ideas and actually develop prototypes, which can help identify problems with a project or shine light on possible new features and interactions. As programming becomes more and more popular between people of fields not related to software there is the need for guidance in developing something other than basic algorithms, which is where this thesis comes in: A comprehensive document explaining the challenges and available choices of developing a sensor data and message delivery system, which scales well and implements the delivery of critical messages. Modularity and extensibility were also given much importance, making this an affordable tool for anyone that wants to build a sensor network of the kind.

The integration of online advertising: The effects of source expertise and certainty level of the message on consumer involvement and persuasion in the context of user-generated advertising

A Work Project, presented as part of the requirements for the Award of a Masters Degree in Management from the NOVA – School of Business and Economics

Personalization platform for multimodal ubiquitous computing applications

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Fatores de prognóstico para os resultados de sucesso da fisioterapia multimodal em utentes com dor cervical crónica

RESUMO: Introdução: O conhecimento acerca da influência das características dos indivíduos com dor cervical crónica (DCC) no prognóstico dos resultados alcançados com a Fisioterapia é ainda inconsistente, sendo escassos os estudos desenvolvidos neste âmbito. Objetivo: Este relatório pretende determinar se um modelo baseado em fatores de prognóstico é capaz de prever os resultados de sucesso da Fisioterapia, a curto prazo, em utentes com DCC, ao nível da incapacidade funcional, intensidade da dor e perceção global de melhoria. Metodologia: Trata-se de estudo de coorte prospetivo com 112 participantes. Os utentes foram avaliados na primeira semana de tratamento e sete semanas após o início da intervenção. Os instrumentos utilizados foram o Neck Disability Index–Versão Portuguesa (NDI-PT) e a Escala Numérica da Dor (END) nos dois momentos de avaliação, um Questionário de Caracterização Sociodemográfica e Clínica da Amostra na baseline e a Patient Global Impression Change Scale–Versão Portuguesa (PGIC-PT) no follow-up. As características sociodemográficas e clínicas foram incluídas como potenciais fatores de prognóstico e estes foram definidos com base nas diferenças mínimas clinicamente importantes (DMCI) dos instrumentos NDIPT (DMCI≥6) e END (DMCI≥2) e no critério de pontuação ≥5 na PGIC-PT. A análise dos dados foi realizada através do método de regressão logística (backward conditional procedure) para identificar as associações entre os indicadores e as variáveis de resultado (p<0.05). Resultados: Dos 112 participantes incluídos no estudo, 108 completaram o follow-up (média de idade: 51.76±10.19). No modelo multivariado relativo à incapacidade funcional, os resultados de sucesso encontram-se associados a elevados níveis de incapacidade na baseline (OR=1.123; 95% IC 1.056–1.194) e a duração da dor inferior a 12 meses (OR=2.704; 95% IC 1.138–6.424). Este modelo explica 30.0% da variância da melhoria da funcionalidade e classifica corretamente 74.1% dos utentes (sensibilidade: 75.9%; especificidade: 72.0%). O modelo relativo à intensidade da dor identificou apenas a associação do outcome com níveis elevados de intensidade da dor na baseline (OR=1.321; 95% IC 1.047–1.668), explicando 7.5% da variância da redução da mesma e classificando corretamente 68.2% dos utentes (sensibilidade: 94.4%; especificidade: 16.7%). O modelo final referente à perceção global de melhoria apresentou uma associação com a intensidade da dor na baseline (OR=0.621; 95% IC 0.465–0.829), com a presença de cefaleias e/ou tonturas (OR=2.538; 95% IC 0.987–6.526) e com a duração da dor superior a 12 meses (OR=0.279; 95% IC 0.109–0.719). Este modelo explica 27.5% da variância dos resultados de sucesso para este outcome e classifica corretamente 73.1% dos utentes (sensibilidade: 81.8%; especificidade: 59.5%). Conclusões: Utentes com DCC com elevada incapacidade na baseline e queixas de dor há menos de 12 meses apresentam maior probabilidade de obter melhorias ao nível da incapacidade funcional. Elevados níveis de intensidade da dor na baseline predizem resultados de sucesso na redução da dor após sete semanas de tratamento. Utentes com DCC com baixos níveis de dor na baseline, com cefaleias e/ou tonturas e com queixas de dor há mais de 12 meses apresentam maior probabilidade de obter uma melhor perceção de melhoria.--------------- ABSTRACT:Introduction: The influence of the characteristics of individuals with chronic neck pain (CNP) on the prognosis of physiotherapy outcomes is still inconsistent, there being few studies developed in this context. Aim: This study seeks to determine whether a model based on prognostic factors can predict the short-term physiotherapy successful outcomes in CNP patients, regarding functional disability, pain intensity and perceived recovery. Methodology: This is a prospective cohort study with 112 participants. Patients were assessed during the first week of treatment and seven weeks after the start of the intervention. The instruments used were the Neck Disability Index–Portuguese Version (NDI-PT) and the Numerical Rating Scale (NRS) at both moments of assessment, a Sample Sociodemographic and Clinical Characterization Questionnaire at baseline and Patient Global Impression Change Scale–Portuguese Version (PGIC-PT) at the follow-up. The sociodemographic and clinical characteristics were included as potential predictors of successful outcomes, and these were defined on the basis of minimal clinically important differences (MCID) of NDI-PT (MCID≥6) and END (MCID≥2) and the criteria score ≥5 on the PGIC-PT. Data analysis was performed using logistic regression (backward conditional procedure) to identify associations between predictors and outcomes (p<0.05). Results: Of the 112 participants included in the study, 108 completed the follow-up (mean age: 51.76±10.19). In the multivariate model of functional disability, the successful outcomes are associated with high levels of disability at baseline (OR = 1.123; 95% CI 1.056-1.194), and pain duration shorter than 12 months (OR=2.704; 95% CI 1.138–6.424). This model explains 30.0% of the variance of improved functional capacity and correctly classifies 74.1% of the patients (sensitivity: 75.9%, specificity: 72.0%). The model for pain intensity solely identified an outcome association with high pain intensity at baseline (OR=1.321; 95% CI 1.047-1.668), explaining 7.5% of the variance of pain reduction and correctly classifying 68.2% of the patients (sensitivity: 94.4%, specificity: 16.7%). The final model of perceived recovery showed an association with pain intensity at baseline (OR=0.621; 95% CI 0465-0829), with the presence of headache and/or dizziness (OR=2.538; 95% CI 0.987-6.526) and the duration of pain over 12 months (OR=0.279; 95% CI 0.109-0.719). This model explains 27.5% of the variance of successful outcomes and correctly classifies 73.1% of the patients (sensitivity: 81.8%, specificity: 59.5%). Conclusions: Patients with CNP with high disability at baseline and complaints of pain for less than 12 months are more likely to obtain improvements in functional disability. High levels of pain intensity at baseline predict successful outcomes in pain reduction after seven weeks of treatment. Patients with CNP with low levels of pain at baseline, with headache and/or dizziness and with pain complaints for more than 12 months are more likely to get a better perceived recovery.