Estimation of State Space Models and Stochastic Volatility

Ma thèse est composée de trois chapitres reliés à l'estimation des modèles espace-état et volatilité stochastique. Dans le première article, nous développons une procédure de lissage de l'état, avec efficacité computationnelle, dans un modèle espace-état linéaire et gaussien. Nous montrons comment exploiter la structure particulière des modèles espace-état pour tirer les états latents efficacement. Nous analysons l'efficacité computationnelle des méthodes basées sur le filtre de Kalman, l'algorithme facteur de Cholesky et notre nouvelle méthode utilisant le compte d'opérations et d'expériences de calcul. Nous montrons que pour de nombreux cas importants, notre méthode est plus efficace. Les gains sont particulièrement grands pour les cas où la dimension des variables observées est grande ou dans les cas où il faut faire des tirages répétés des états pour les mêmes valeurs de paramètres. Comme application, on considère un modèle multivarié de Poisson avec le temps des intensités variables, lequel est utilisé pour analyser le compte de données des transactions sur les marchés financières. Dans le deuxième chapitre, nous proposons une nouvelle technique pour analyser des modèles multivariés à volatilité stochastique. La méthode proposée est basée sur le tirage efficace de la volatilité de son densité conditionnelle sachant les paramètres et les données. Notre méthodologie s'applique aux modèles avec plusieurs types de dépendance dans la coupe transversale. Nous pouvons modeler des matrices de corrélation conditionnelles variant dans le temps en incorporant des facteurs dans l'équation de rendements, où les facteurs sont des processus de volatilité stochastique indépendants. Nous pouvons incorporer des copules pour permettre la dépendance conditionnelle des rendements sachant la volatilité, permettant avoir différent lois marginaux de Student avec des degrés de liberté spécifiques pour capturer l'hétérogénéité des rendements. On tire la volatilité comme un bloc dans la dimension du temps et un à la fois dans la dimension de la coupe transversale. Nous appliquons la méthode introduite par McCausland (2012) pour obtenir une bonne approximation de la distribution conditionnelle à posteriori de la volatilité d'un rendement sachant les volatilités d'autres rendements, les paramètres et les corrélations dynamiques. Le modèle est évalué en utilisant des données réelles pour dix taux de change. Nous rapportons des résultats pour des modèles univariés de volatilité stochastique et deux modèles multivariés. Dans le troisième chapitre, nous évaluons l'information contribuée par des variations de volatilite réalisée à l'évaluation et prévision de la volatilité quand des prix sont mesurés avec et sans erreur. Nous utilisons de modèles de volatilité stochastique. Nous considérons le point de vue d'un investisseur pour qui la volatilité est une variable latent inconnu et la volatilité réalisée est une quantité d'échantillon qui contient des informations sur lui. Nous employons des méthodes bayésiennes de Monte Carlo par chaîne de Markov pour estimer les modèles, qui permettent la formulation, non seulement des densités a posteriori de la volatilité, mais aussi les densités prédictives de la volatilité future. Nous comparons les prévisions de volatilité et les taux de succès des prévisions qui emploient et n'emploient pas l'information contenue dans la volatilité réalisée. Cette approche se distingue de celles existantes dans la littérature empirique en ce sens que ces dernières se limitent le plus souvent à documenter la capacité de la volatilité réalisée à se prévoir à elle-même. Nous présentons des applications empiriques en utilisant les rendements journaliers des indices et de taux de change. Les différents modèles concurrents sont appliqués à la seconde moitié de 2008, une période marquante dans la récente crise financière.

Squared interaction matrix Sherrington-Kirkpatrick model for a spin glass

The mean-field theory of a spin glass with a specific form of nearest- and next-nearest-neighbor interactions is investigated. Depending on the sign of the interaction matrix chosen we find either the continuous replica symmetry breaking seen in the Sherrington-Kirkpartick model or a one-step solution similar to that found in structural glasses. Our results are confirmed by numerical simulations and the link between the type of spin-glass behavior and the density of eigenvalues of the interaction matrix is discussed.

Modified Block Adaptive Predictive Coder For Speech Processing

This thesis investigated the potential use of Linear Predictive Coding in speech communication applications. A Modified Block Adaptive Predictive Coder is developed, which reduces the computational burden and complexity without sacrificing the speech quality, as compared to the conventional adaptive predictive coding (APC) system. For this, changes in the evaluation methods have been evolved. This method is as different from the usual APC system in that the difference between the true and the predicted value is not transmitted. This allows the replacement of the high order predictor in the transmitter section of a predictive coding system, by a simple delay unit, which makes the transmitter quite simple. Also, the block length used in the processing of the speech signal is adjusted relative to the pitch period of the signal being processed rather than choosing a constant length as hitherto done by other researchers. The efficiency of the newly proposed coder has been supported with results of computer simulation using real speech data. Three methods for voiced/unvoiced/silent/transition classification have been presented. The first one is based on energy, zerocrossing rate and the periodicity of the waveform. The second method uses normalised correlation coefficient as the main parameter, while the third method utilizes a pitch-dependent correlation factor. The third algorithm which gives the minimum error probability has been chosen in a later chapter to design the modified coder The thesis also presents a comparazive study beh-cm the autocorrelation and the covariance methods used in the evaluaiicn of the predictor parameters. It has been proved that the azztocorrelation method is superior to the covariance method with respect to the filter stabf-it)‘ and also in an SNR sense, though the increase in gain is only small. The Modified Block Adaptive Coder applies a switching from pitch precitzion to spectrum prediction when the speech segment changes from a voiced or transition region to an unvoiced region. The experiments cont;-:ted in coding, transmission and simulation, used speech samples from .\£=_‘ajr2_1a:r1 and English phrases. Proposal for a speaker reecgnifion syste: and a phoneme identification system has also been outlized towards the end of the thesis.

The sediment trapping efficiency of the macro-tidal Daly Estuary, tropical Australia

Field studies were carried out on the water and sediment dynamics in the tropical, macro-tidal, Daly Estuary. The estuary is shallow, very-turbid, about 100 km long, and the entrance is funnel-shape. In the wet, high flow season, normal tidal ranges can be suppressed in the estuary, depending on inflow rates, and freshwater becomes dominant up to the mouth. At that time a fraction of the fine sediment load is exported offshore as a bottom-tagging nepheloid layer after the sediment falls out of suspension of the thin, near-surface, river plume. The remaining fraction and the riverine coarse sediment form a large sediment bar 10 km long, up to 6 m in height and extending across the whole width of the channel near the mouth. This bar, as well as shoals in the estuary, partially pond the mid- to upper-estuary. This bar builds up from the deposition of riverine sediment during a wet season with high runoff and can raise mean water level by up to 2 m in the upper estuary in the low flow season. This ponding effect takes about three successive dry years to disappear by the sediment forming the bar being redistributed all over the estuary by tidal pumping of fine and coarse sediment in the dry season, which is the low flow season. The swift reversal of the tidal currents from ebb to flood results in macro-turbulence that lasts about 20 min. Bed load transport is preferentially landward and occurs only for water currents greater than 0.6 m s(-1). This high value of the threshold velocity suggests that the sand may be cemented by the mud. The Daly Estuary thus is a leaky sediment trap with an efficiency varying both seasonally and inter-annually. (c) 2006 Elsevier Ltd. All rights reserved.

Profitability and technical efficiency of aquaculture systems in Pampanga, Philippines

We evaluate the profitability and technical efficiency of aquaculture in the Philippines. Farm-level data are used to compare two production systems corresponding to the intensive monoculture of tilapia in freshwater ponds and the extensive polyculture of shrimps and fish in brackish water ponds. Both activities are very lucrative, with brackish water aquaculture achieving the higher level of profit per farm. Stochastic frontier production functions reveal that technical efficiency is low in brackish water aquaculture, with a mean of 53%, explained primarily by the operator's experience and by the frequency of his visits to the farm. In freshwater aquaculture, the farms achieve a mean efficiency level of 83%. The results suggest that the provision of extension services to brackish water fish farms might be a cost-effective way of increasing production and productivity in that sector. By contrast, technological change will have to be the driving force of future productivity growth in freshwater aquaculture.

Automatic nonlinear predictive model-construction algorithm using forward regression and the PRESS statistic

An automatic nonlinear predictive model-construction algorithm is introduced based on forward regression and the predicted-residual-sums-of-squares (PRESS) statistic. The proposed algorithm is based on the fundamental concept of evaluating a model's generalisation capability through crossvalidation. This is achieved by using the PRESS statistic as a cost function to optimise model structure. In particular, the proposed algorithm is developed with the aim of achieving computational efficiency, such that the computational effort, which would usually be extensive in the computation of the PRESS statistic, is reduced or minimised. The computation of PRESS is simplified by avoiding a matrix inversion through the use of the orthogonalisation procedure inherent in forward regression, and is further reduced significantly by the introduction of a forward-recursive formula. Based on the properties of the PRESS statistic, the proposed algorithm can achieve a fully automated procedure without resort to any other validation data set for iterative model evaluation. Numerical examples are used to demonstrate the efficacy of the algorithm.

Prediction of coagulation properties, titratable acidity and pH of bovine milk using mid-infrared spectroscopy

This study investigated the potential application of mid-infrared spectroscopy (MIR 4,000–900 cm−1) for the determination of milk coagulation properties (MCP), titratable acidity (TA), and pH in Brown Swiss milk samples (n = 1,064). Because MCP directly influence the efficiency of the cheese-making process, there is strong industrial interest in developing a rapid method for their assessment. Currently, the determination of MCP involves time-consuming laboratory-based measurements, and it is not feasible to carry out these measurements on the large numbers of milk samples associated with milk recording programs. Mid-infrared spectroscopy is an objective and nondestructive technique providing rapid real-time analysis of food compositional and quality parameters. Analysis of milk rennet coagulation time (RCT, min), curd firmness (a30, mm), TA (SH°/50 mL; SH° = Soxhlet-Henkel degree), and pH was carried out, and MIR data were recorded over the spectral range of 4,000 to 900 cm−1. Models were developed by partial least squares regression using untreated and pretreated spectra. The MCP, TA, and pH prediction models were improved by using the combined spectral ranges of 1,600 to 900 cm−1, 3,040 to 1,700 cm−1, and 4,000 to 3,470 cm−1. The root mean square errors of cross-validation for the developed models were 2.36 min (RCT, range 24.9 min), 6.86 mm (a30, range 58 mm), 0.25 SH°/50 mL (TA, range 3.58 SH°/50 mL), and 0.07 (pH, range 1.15). The most successfully predicted attributes were TA, RCT, and pH. The model for the prediction of TA provided approximate prediction (R2 = 0.66), whereas the predictive models developed for RCT and pH could discriminate between high and low values (R2 = 0.59 to 0.62). It was concluded that, although the models require further development to improve their accuracy before their application in industry, MIR spectroscopy has potential application for the assessment of RCT, TA, and pH during routine milk analysis in the dairy industry. The implementation of such models could be a means of improving MCP through phenotypic-based selection programs and to amend milk payment systems to incorporate MCP into their payment criteria.

Efficiency of pseudo-spectral algorithms with Anderson mixing for the SCFT of periodic block-copolymer phases

This study examines the numerical accuracy, computational cost, and memory requirements of self-consistent field theory (SCFT) calculations when the diffusion equations are solved with various pseudo-spectral methods and the mean field equations are iterated with Anderson mixing. The different methods are tested on the triply-periodic gyroid and spherical phases of a diblock-copolymer melt over a range of intermediate segregations. Anderson mixing is found to be somewhat less effective than when combined with the full-spectral method, but it nevertheless functions admirably well provided that a large number of histories is used. Of the different pseudo-spectral algorithms, the 4th-order one of Ranjan, Qin and Morse performs best, although not quite as efficiently as the full-spectral method.

Using rain-use efficiency to explore livestock production trends in rangelands in the Transkei, South Africa

We qualitatively describe the condition of communally managed rangelands in the Transkei, South Africa, using GIS and high resolution near-infrared imagery. Using livestock census data from 28 magisterial districts in the Transkei, we explored the trends in livestock biomass from 1923–1998. The area had been subjected to intensive herbivory by domestic livestock during that period, and the high livestock biomass had been blamed for the perceived degradation or ‘overgrazing’ of the region. Our assessment used the concept rain-use efficiency (RUE) (kg dry matter ha–1 mm–1) to determine whether there is evidence of change in the efficiency of the system to produce domestic livestock. We calculated RUE from annual livestock numbers and the mean annual rainfall for each district. We found no evidence of a decline in rain-use efficiency between the two assessment periods (1923–1944, 1945–1998). There was evidence of a shift in the ratio of sheep to goats between 1923 and 1998, with goat numbers increasing (greater than twofold) relative to sheep in eight districts. This trend may be associated with changes in the structure of vegetation. We conclude that this region is not showing evidence of system run down that affects domestic livestock production.

Does more detailed information mean better performance? An experiment in information explicitness

Purpose – Investors are now able to analyse more noise-free news to inform their trading decisions than ever before. Their expectation that more information means better performance is not supported by previous psychological experiments which argue that too much information actually impairs performance. The purpose of this paper is to examine whether the degree of information explicitness improves stock market performance. Design/methodology/approach – An experiment is conducted in a computer laboratory to examine a trading simulation manipulated from a real market-shock. Participants’ performance efficiency and effectiveness are measured separately. Findings – The results indicate that the explicitness of information neither improves nor impairs participants’ performance effectiveness from the perspectives of returns, share and cash positions, and trading volumes. However, participants’ performance efficiency is significantly affected by information explicitness. Originality/value – The novel approach and findings of this research add to the knowledge of the impact of information explicitness on the quality of decision making in a financial market environment.

Do turning biases by the 7-spot ladybird, Coccinella septempunctata, increase their foraging efficiency?

The hypothesis that foraging male and female Coccinella septempunctata L. would exhibit a turning bias when walking along a branched linear wire in a Y-maze was tested. Individuals were placed repeatedly in the maze. Approximately 45% of all individuals tested displayed significant turning biases, with a similar number of individuals biased to the left and right. In the maze right-handed individuals turned right at 84.4% of turns and the left-handed individuals turned left at 80.2% of turns. A model of the searching efficiency of C. septempunctata in dichotomous branched environments showed that model coccinellids with greater turning biases discovered a higher proportion of the plant for a given number of searches than those with no bias. A modification of the model to investigate foraging efficiency, by calculating the mean time taken by individuals to find randomly distributed aphid patches, suggested that on four different sizes of plants, with a variety of aphid patch densities, implementing a turning bias was a significantly more efficient foraging strategy than no bias. In general the benefits to foraging of implementing a turning bias increased with the degree of the bias. It may be beneficial for individuals in highly complex branched environments to have a turning bias slightly lower than 100% in order to benefit from increased foraging efficiency without walking in circles. Foraging bias benefits increased with increasing plant size and decreasing aphid density. In comparisons of two different plant morphologies, one with a straight stem and side branches and one with a symmetrically branched morphology, there were few significant differences in the effects of turning biases on foraging efficiency between morphologies

Efficiency of uniparental male and female care against egg predators in two closely related syntopic harvestmen

Although the benefits of maternal care have been investigated in many species, the caring role of males in species with exclusive paternal care has received less attention. We experimentally quantified the protective role of paternal care in the harvestman Iporangaia pustulosa. Additionally, we compared the effectiveness of paternal care against predation in this species with a syntopic harvestman with maternal care, Acutisoma proximum. We demonstrated that nearly one-third of the unprotected Iporangaia clutches disappeared entirely in 12 days, while the other two-thirds suffered a mean reduction of 55% in egg number. Conversely, 50% of the control clutches did not suffer any reduction, and only one was entirely consumed by predators. We also demonstrated that the mucus coat that covers Iporangaia clutches has an important deterrent role against predation by conspecifics: 58.3% of the clutches without mucus were attacked and three of them were entirely consumed, whereas only three clutches with mucus were attacked, suffering a reduction of up to three eggs. Iporangaia males were as efficient as Acutisoma females in protecting eggs. However, unattended Acutisoma eggs were attacked 20% more frequently than unattended Iporangaia eggs. Unattended Iporangaia eggs are protected by a mucus coat that prevents or decreases predation rate, whereas Acutisoma eggs are more susceptible to predation, probably because they lack this mucus coat. Thus, besides the fact that Iporangaia males efficiently protect the offspring against egg predators, females also contribute to egg protection by providing a mucus coat that deters egg predators. (C) 2009 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Predictive models for chronic renal disease using decision trees, naïve bayes and case-based methods

Data mining can be used in healthcare industry to “mine” clinical data to discover hidden information for intelligent and affective decision making. Discovery of hidden patterns and relationships often goes intact, yet advanced data mining techniques can be helpful as remedy to this scenario. This thesis mainly deals with Intelligent Prediction of Chronic Renal Disease (IPCRD). Data covers blood, urine test, and external symptoms applied to predict chronic renal disease. Data from the database is initially transformed to Weka (3.6) and Chi-Square method is used for features section. After normalizing data, three classifiers were applied and efficiency of output is evaluated. Mainly, three classifiers are analyzed: Decision Tree, Naïve Bayes, K-Nearest Neighbour algorithm. Results show that each technique has its unique strength in realizing the objectives of the defined mining goals. Efficiency of Decision Tree and KNN was almost same but Naïve Bayes proved a comparative edge over others. Further sensitivity and specificity tests are used as statistical measures to examine the performance of a binary classification. Sensitivity (also called recall rate in some fields) measures the proportion of actual positives which are correctly identified while Specificity measures the proportion of negatives which are correctly identified. CRISP-DM methodology is applied to build the mining models. It consists of six major phases: business understanding, data understanding, data preparation, modeling, evaluation, and deployment.

Incidência de infecção por Legionella pneumophila em pacientes que internaram no HCPA com pneumonia adquirida na comunidade

Introdução: O diagnóstico microbiológico da infecção por Legionella é complexo, pois a bactéria não é visualizada à coloração de Gram no escarro, e sua cultura não é realizada na maioria dos laboratórios clínicos. A imunofluorescência direta nas secreções respiratórias tem baixa sensibilidade, em torno de 40% e a técnica da “PCR” não é ainda recomendada para o diagnóstico clínico (CDC, 1997). A detecção de anticorpos no soro é a técnica mais utilizada, e o critério definitivo é a soroconversão para no mínimo 1:128, cuja sensibilidade é de 70 a 80% (Edelstein, 1993). Como critérios diagnósticos de possível pneumonia por Legionella, eram utilizados: título único de anticorpos a L pneumophila positivo na diluição 1:256, em paciente com quadro clínico compatível (CDC, 1990) e o achado de antígeno a Legionella na urina (WHO, 1990). Nos últimos anos, porém, com o uso crescente do teste de antigenúria, foram detectados casos de pneumonia por Legionella, que não eram diagnosticados por cultura ou sorologia, tornando-o método diagnóstico de certeza para o diagnóstico de pneumonia por Legionella (CDC, 1997). Por sua fácil execução, resultado imediato, e alta sensibilidade - de 86% a 98% (Kashuba & Ballow, 1986; Harrison & Doshi, 2001), tem sido recomendado para o diagnóstico das PAC que necessitam internação hospitalar (Mulazimoglu & Yu, 2001; Gupta et al., 2001; Marrie, 2001), especialmente em UTI (ATS, 2001). Vários estudos documentaram baixo valor preditivo positivo do título único positivo de 1:256, tornando-o sem valor para o diagnóstico da pneumonia por Legionella, exceto, talvez, em surtos (Plouffe et al., 1995). Outros detectaram alta prevalência de anticorpos positivos na diluição 1:256 na população, em pessoas normais (Wilkinson et al., 1983; Nichol et al., 1991). A partir de 1996, o CDC de Atlanta recomendou que não seja mais utilizado o critério de caso provável de infecção por Legionella pneumophila por título único de fase convalescente ≥1:256, por falta de especificidade(CDC, 1997). A pneumonia por Legionella é raramente diagnosticada, e sua incidência é subestimada. Em estudos de PAC, a incidência da pneumonia por Legionella nos EUA, Europa, Israel e Austrália, foi estimada entre 1% a 16% (Muder & Yu, 2000). Nos EUA, foi estimado que cerca de 8 000 a 23 000 casos de PAC por Legionella ocorrem anualmente, em pacientes que requerem hospitalização (Marston et al., 1994 e 1977). No Brasil, a incidência de PAC causadas por Legionella em pacientes hospitalizados é tema de investigação pertinente, ainda não relatado na literatura. Objetivo: detectar a incidência de pneumonias causadas por Legionella pneumophila sorogrupos 1 a 6, em pacientes que internaram no Hospital de Clínicas de Porto Alegre por PAC, por um ano. Material e Métodos: o delineamento escolhido foi um estudo de coorte (de incidência), constituída por casos consecutivos de pneumonia adquirida na comunidade que internaram no HCPA de 19 de julho de 2000 a 18 de julho de 2001. Para a identificação dos casos, foram examinados diariamente o registro computadorizado das internações hospitalares, exceto as internações da pediatria e da obstetrícia, sendo selecionados todos os pacientes internados com o diagnóstico de pneumonia e de insuficiência respiratória aguda. Foram excluídos aqueles com menos de 18 anos ou mais de 80 anos; os procedentes de instituições, HIV-positivos, gestantes, pacientes restritos ao leito; e portadores de doença estrutural pulmonar ou traqueostomias. Foram excluídos os pacientes que tivessem tido alta hospitalar nos últimos 15 dias, e aqueles já incluídos no decorrer do estudo. Os pacientes selecionados foram examinados por um pesquisador, e incluídos para estudo se apresentassem infiltrado ao RX de tórax compatível com pneumonia, associado a pelo menos um dos sintomas respiratórios maiores (temperatura axilar > 37,8ºC, tosse ou escarro; ou dois sintomas menores (pleurisia, dispnéia, alteração do estado mental, sinais de consolidação à ausculta pulmonar, mais de 12 000 leucócitos/mm3). O estudo foi previamente aprovado pela Comissão de Ética em Pesquisa do HCPA. Os pacientes eram entrevistados por um pesquisador, dando seu consentimento por escrito, e então seus dados clínicos e laboratoriais eram registrados em protocolo individual. Não houve interferência do pesquisador, durante a internação, exceto pela coleta de urina e de sangue para exame laboratoriais específicos da pesquisa. Os pacientes eram agendados, no ambulatório de pesquisa, num prazo de 4 a 12 semanas após sua inclusão no estudo, quando realizavam nova coleta de sangue, RX de tórax de controle, e outros exames que se fizessem necessários para esclarecimento diagnóstico.Todos os pacientes foram acompanhados por 1 ano, após sua inclusão no estudo.Foram utilizadas a técnica de imunofluorescência indireta para detecção de anticorpos das classes IgG, IgM e IgA a Legionella pneumophila sorogrupos 1 a 6 no soro, em duas amostras, colhidas, respectivamente, na 1ª semana de internação e depois de 4 a 12 semanas; e a técnica imunológica por teste ELISA para a detecção do antígeno de Legionella pneumophila sorogrupo 1 na urina, colhida na primeira semana de internação. As urinas eram armazenadas, imediatamente após sua coleta, em freezer a –70ºC, e depois descongeladas e processadas em grupos de cerca de 20 amostras. A imunofluorescência foi feita no laboratório de doenças Infecciosas da Universidade de Louisville (KY, EUA), em amostras de soro da fase aguda e convalescente, a partir da diluição 1:8; e a detecção do antígeno de Legionella pneumophila sorogrupo 1, nas amostras de urina, foi realizada no laboratório de pesquisa do HCPA, pelos investigadores, utilizando um kit comercial de teste ELISA fabricado por Binax (Binax Legionella Urinary Enzyme Assay, Raritan, EUA). As urinas positivas eram recongeladas novamente, para serem enviadas para confirmação no mesmo laboratório americano, ao fim do estudo. Foram adotados como critérios definitivos de infecção por Legionella pneumophila sorogrupos 1 a 6, a soroconversão (elevação de 4 vezes no título de anticorpos séricos entre o soro da fase aguda e da fase convalescente para no mínimo 1:128); ou o achado de antígeno de L pneumophila sorogrupo 1 na urina não concentrada, numa razão superior a 3, conforme instruções do fabricante e da literatura.Os pacientes foram classificados, de acordo com suas características clínicas, em 1º) portadores de doenças crônicas (doenças pulmonares, cardíacas, diabete mellitus, hepatopatias e insuficiência renal); 2º) portadores de doenças subjacentes com imunossupressão; 3º) pacientes hígidos ou com outras doenças que não determinassem insuficiência orgânica. Imunossupressão foi definida como esplenectomia, ser portador de neoplasia hematológica, portador de doença auto-imune, ou de transplante; ou uso de medicação imunossupressora nas 4 semanas anteriores ao diagnóstico (Yu et al., 2002b); ou uso de prednisolona 10 mg/dia ou equivalente nos últimos 3 meses (Lim et al., 2001). As características clínicas e laboratoriais dos pacientes que evoluíram ao óbito por pneumonia foram comparados àquelas dos pacientes que obtiveram cura. Para a análise das variáveis categóricas, utilizou-se o teste qui-quadrado de Pearson ou teste exato de Fisher. Para as variáveis numéricas contínuas, utilizou-se o teste “t“ de Student. Um valor de p< 0,05 foi considerado como resultado estatisticamente significativo (programas SPSS, versão 10). Foi calculada a freqüência de mortes por pneumonia na população estudada, adotando-se a alta hospitalar como critério de cura. Foi calculada a incidência cumulativa para pneumonia por Legionella pneumophila sorogrupos 1 a 6, em um hospital geral, no período de 1 ano. Resultados: durante um ano de estudo foram examinados 645 registros de internação, nos quais constavam, como motivo de baixa hospitalar, o diagnóstico de pneumonia ou de insuficiência respiratória aguda; a maioria desses diagnósticos iniciais não foram confirmados. Desses 645 pacientes, foram incluídos no estudo 82 pacientes, nos quais os critérios clínicos ou radiológicos de pneumonia foram confirmados pelos pesquisadores. Durante o acompanhamento desses pacientes, porém, foram excluídos 23 pacientes por apresentarem outras patologias que mimetizavam pneumonia: DPOC agudizado (5), insuficiência cardíaca (3), tuberculose pulmonar (2), colagenose (1), fibrose pulmonar idiopática (1), edema pulmonar em paciente com cirrose (1), somente infecçâo respiratória em paciente com sequelas pulmonares (4); ou por apresentarem critérios de exclusão: bronquiectasias (4), HIV positivo (1), pneumatocele prévia (1). Ao final, foram estudados 59 pacientes com pneumonia adquirida na comunidade, sendo 20 do sexo feminino e 39 do sexo masculino, com idade entre 24 e 80 anos (média de 57,6 anos e desvio padrão de ±10,6). Tivemos 36 pacientes com doenças subjacentes classificadas como “doenças crônicas”, dos quais 18 pacientes apresentavam mais de uma co-morbidade, por ordem de prevalência: doenças pulmonares, cardíacas, diabete mellitus, hepatopatias e insuficiência renal; neoplasias ocorreram em 9 pacientes, sendo sólidas em 7 pacientes e hematológicas em 2. Dos 59 pacientes, 61% eram tabagistas e 16,9%, alcoolistas. Do total, 10 pacientes apresentavam imunossupressão. Dos demais 13 pacientes, somente um era previamente hígido, enquanto os outros apresentavam tabagismo, sinusite, anemia, HAS, gota, ou arterite de Takayasu. A apresentação radiológica inicial foi broncopneumonia em 59,3% dos casos; pneumonia alveolar ocorreu em 23,7% dos casos, enquanto ambos padrões ocorreram em 15,2% dos pacientes. Pneumonia intersticial ocorreu em somente um caso, enquanto broncopneumonia obstrutiva ocorreu em 5 pacientes (8,5%). Derrame pleural ocorreu em 22% dos casos, e em 21 pacientes (35%) houve comprometimento de mais de um lobo ao RX de tórax. Foram usados beta-lactâmicos para o tratamento da maioria dos pacientes (72,9%9). A segunda classe de antibióticos mais usados foi a das fluoroquinolonas respiratórias, que foram receitadas para 23 pacientes (39,0%), e em 3º lugar, os macrolídeos, usados por 11 pacientes (18,6%). Apenas 16 pacientes não usaram beta-lactâmicos, em sua maioria recebendo quinolonas ou macrolídeos. Dos 43 pacientes que usaram beta-lactâmicos, 25 não usaram nem macrolídeos, nem quinolonas. Em 13 pacientes as fluoroquinolonas respiratórias foram as únicas drogas usadas para o tratamento da pneumonia. Do total, 8 pacientes foram a óbito por pneumonia; em outros 3 pacientes, o óbito foi atribuído a neoplasia em estágio avançado. Dos 48 pacientes que obtiveram cura, 33 (68,7%) estavam vivos após 12 meses. Os resultados da comparação realizada evidenciaram tendência a maior mortalidade no sexo masculino e em pacientes com imunossupressão, porém essa associação não alcançou significância estatística. Os pacientes que usaram somente beta-lactâmicos não apresentaram maior mortalidade do que os pacientes que usaram beta-lactâmicos associados a outras classes de antibióticos ou somente outras classes de antibióticos. Examinando-se os pacientes que utiizaram macrolídeos ou quinolonas em seu regime de tratamento, isoladamente ou combinados a outros antibióticos, observou-se que também não houve diferença dos outros pacientes, quanto à mortalidade. Os pacientes com padrão radiológico de pneumonia alveolar tiveram maior mortalidade, e essa diferença apresentou uma significância limítrofe (p= 0,05). Nossa mortalidade (11,9%) foi similar à de Fang et al. (1990), em estudo clássico de 1991 (13,7%); foi também similar à média de mortalidade das PAC internadas não em UTI (12%), relatada pela ATS, no seu último consenso para o tratamento empírico das PAC (ATS, 2001). Foram detectados 3 pacientes com pneumonia por Legionella pneumophila sorogrupo 1 na população estudada: 2 foram diagnosticados por soroconversão e por antigenúria positiva, e o 3º foi diagnosticado somente pelo critério de antigenúria positiva, tendo sorologia negativa, como alguns autores (McWhinney et al., 2000). Dois pacientes com PAC por Legionella não responderam ao tratamento inicial com beta-lactâmicos, obtendo cura com levofloxacina; o 3º paciente foi tratado somente com betalactâmicos, obtendo cura. Conclusões: A incidência anual de PAC por Legionella pneumophila sorogrupos 1 a 6, no HCPA, foi de 5,1%, que representa a incidência anual de PAC por Legionella pneumophila sorogrupos 1 a 6 em um hospital geral universitário. Comentários e Perspectivas: Há necessidade de se empregar métodos diagnósticos específicos para o diagnóstico das pneumonias por Legionella em nosso meio, como a cultura, a sorologia com detecção de todas as classes de anticorpos, e a detecção do antígeno urinário, pois somente com o uso simultâneo de técnicas complementares pode-se detectar a incidência real de pneumonias causadas tanto por Legionella pneumophila, como por outras espécies. A detecção do antígeno de Legionella na urina é o teste diagnóstico de maior rendimento, sendo recomendado seu uso em todas as PAC que necessitarem internação hospitalar (Mulazimoglu & Yu, 2001; Gupta et al., 2001); em todos os pacientes com PAC que apresentarem fatores de risco potenciais para legionelose (Marrie, 2001); e para o diagnóstico etiológico das pneumonias graves (ATS, 2001). Seu uso é indicado, com unanimidade na literatura, para a pesquisa de legionelose nosocomial e de surtos de legionelose na comunidade.

Forecasting Brazilian inflation by its aggregate and disaggregated data: a test of predictive power by forecast horizon

This work aims to compare the forecast efficiency of different types of methodologies applied to Brazilian Consumer inflation (IPCA). We will compare forecasting models using disaggregated and aggregated data over twelve months ahead. The disaggregated models were estimated by SARIMA and will have different levels of disaggregation. Aggregated models will be estimated by time series techniques such as SARIMA, state-space structural models and Markov-switching. The forecasting accuracy comparison will be made by the selection model procedure known as Model Confidence Set and by Diebold-Mariano procedure. We were able to find evidence of forecast accuracy gains in models using more disaggregated data