Non-linear time series models

The last three decades have seen quite dramatic changes the way we modeled time dependent data. Linear processes have been in the center stage in modeling time series. As far as the second order properties are concerned, the theory and the methodology are very adequate.However, there are more and more evidences that linear models are not sufficiently flexible and rich enough for modeling purposes and that failure to account for non-linearities can be very misleading and have undesired consequences.

A score test for non-nested hypotheses with applications to discrete data models

This paper suggests that a convenient score test against non-nested alternatives can be constructed from the linear combination of the likelihood functions of the competing models. It is shown that this procedure is essentially a test for the correct specification of the conditional distribution of the variable of interest.

Longitudinal analysis of tumor marker CEA of breast cancer patients from Braga's hospital

Allied to an epidemiological study of population of the Senology Unit of Braga’s Hospital that have been diagnosed with malignant breast cancer, we describe the progression in time of repeated measurements of tumor marker Carcinoembryonic antigen (CEA). Our main purpose is to describe the progression of this tumor marker as a function of possible risk factors and, hence, to understand how these risk factors influences that progression. The response variable, values of CEA, was analyzed making use of longitudinal models, testing for different correlation structures. The same covariates used in a previous survival analysis were considered in the longitudinal model. The reference time used was time from diagnose until death from breast cancer. For diagnostic of the models fitted we have used empirical and theoretical variograms. To evaluate the fixed term of the longitudinal model we have tested for a changing point on the effect of time on the tumor marker progression. A longitudinal model was also fitted only to the subset of patients that died from breast cancer, using the reference time as time from date of death until blood test.

Performance of state space and ARIMA models for consumer retail sales forecasting

Forecasting future sales is one of the most important issues that is beyond all strategic and planning decisions in effective operations of retail businesses. For profitable retail businesses, accurate demand forecasting is crucial in organizing and planning production, purchasing, transportation and labor force. Retail sales series belong to a special type of time series that typically contain trend and seasonal patterns, presenting challenges in developing effective forecasting models. This work compares the forecasting performance of state space models and ARIMA models. The forecasting performance is demonstrated through a case study of retail sales of five different categories of women footwear: Boots, Booties, Flats, Sandals and Shoes. On both methodologies the model with the minimum value of Akaike's Information Criteria for the in-sample period was selected from all admissible models for further evaluation in the out-of-sample. Both one-step and multiple-step forecasts were produced. The results show that when an automatic algorithm the overall out-of-sample forecasting performance of state space and ARIMA models evaluated via RMSE, MAE and MAPE is quite similar on both one-step and multi-step forecasts. We also conclude that state space and ARIMA produce coverage probabilities that are close to the nominal rates for both one-step and multi-step forecasts.

Tuple-based morphisms for interoperability establishment of financial information models

Dissertação apresentada na Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para obtenção do grau de Mestre em Engenharia Electrotécnica e Computadores

Modeling in-Hospital Patient Survival During the First 28 Days After Intensive Care Unit Admission: a Prognostic Model for Clinical Trials in General Critically Ill Patients

OBJECTIVE: The objective of the study was to develop a model for estimating patient 28-day in-hospital mortality using 2 different statistical approaches. DESIGN: The study was designed to develop an outcome prediction model for 28-day in-hospital mortality using (a) logistic regression with random effects and (b) a multilevel Cox proportional hazards model. SETTING: The study involved 305 intensive care units (ICUs) from the basic Simplified Acute Physiology Score (SAPS) 3 cohort. PATIENTS AND PARTICIPANTS: Patients (n = 17138) were from the SAPS 3 database with follow-up data pertaining to the first 28 days in hospital after ICU admission. INTERVENTIONS: None. MEASUREMENTS AND RESULTS: The database was divided randomly into 5 roughly equal-sized parts (at the ICU level). It was thus possible to run the model-building procedure 5 times, each time taking four fifths of the sample as a development set and the remaining fifth as the validation set. At 28 days after ICU admission, 19.98% of the patients were still in the hospital. Because of the different sampling space and outcome variables, both models presented a better fit in this sample than did the SAPS 3 admission score calibrated to vital status at hospital discharge, both on the general population and in major subgroups. CONCLUSIONS: Both statistical methods can be used to model the 28-day in-hospital mortality better than the SAPS 3 admission model. However, because the logistic regression approach is specifically designed to forecast 28-day mortality, and given the high uncertainty associated with the assumption of the proportionality of risks in the Cox model, the logistic regression approach proved to be superior.

Characterizing the Risk Profiles of Intensive Care Units

OBJECTIVE: To develop a new method to evaluate the performance of individual ICUs through the calculation and visualisation of risk profiles. METHODS: The study included 102,561 patients consecutively admitted to 77 ICUs in Austria. We customized the function which predicts hospital mortality (using SAPS II) for each ICU. We then compared the risks of hospital mortality resulting from this function with the risks which would be obtained using the original function. The derived risk ratio was then plotted together with point-wise confidence intervals in order to visualise the individual risk profile of each ICU over the whole spectrum of expected hospital mortality. MAIN MEASUREMENTS AND RESULTS: We calculated risk profiles for all ICUs in the ASDI data set according to the proposed method. We show examples how the clinical performance of ICUs may depend on the severity of illness of their patients. Both the distribution of the Hosmer-Lemeshow goodness-of-fit test statistics and the histogram of the corresponding P values demonstrated a good fit of the individual risk models. CONCLUSIONS: Our risk profile model makes it possible to evaluate ICUs on the basis of the specific risk for patients to die compared to a reference sample over the whole spectrum of hospital mortality. Thus, ICUs at different levels of severity of illness can be directly compared, giving a clear advantage over the use of the conventional single point estimate of the overall observed-to-expected mortality ratio.

ANIMAL MODELS FOR THE STUDY OF LEISHMANIASIS IMMUNOLOGY

Leishmaniasis remains a major public health problem worldwide and is classified as Category I by the TDR/WHO, mainly due to the absence of control. Many experimental models like rodents, dogs and monkeys have been developed, each with specific features, in order to characterize the immune response to Leishmania species, but none reproduces the pathology observed in human disease. Conflicting data may arise in part because different parasite strains or species are being examined, different tissue targets (mice footpad, ear, or base of tail) are being infected, and different numbers (“low” 1×102 and “high” 1×106) of metacyclic promastigotes have been inoculated. Recently, new approaches have been proposed to provide more meaningful data regarding the host response and pathogenesis that parallels human disease. The use of sand fly saliva and low numbers of parasites in experimental infections has led to mimic natural transmission and find new molecules and immune mechanisms which should be considered when designing vaccines and control strategies. Moreover, the use of wild rodents as experimental models has been proposed as a good alternative for studying the host-pathogen relationships and for testing candidate vaccines. To date, using natural reservoirs to study Leishmania infection has been challenging because immunologic reagents for use in wild rodents are lacking. This review discusses the principal immunological findings against Leishmania infection in different animal models highlighting the importance of using experimental conditions similar to natural transmission and reservoir species as experimental models to study the immunopathology of the disease.

Landslide susceptibility assessment in Karanganyar regency - Indonesia - Comparison of knowledge-based and Data-driven Models

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Mecanismos de inflamação brônquica resultantes da exposição a factores ambientais

RESUMO Tratando-se a asma de uma doença respiratória, desde há várias décadas que tem sido abordada a hipótese de que factores ambientais, nomeadamente os relacionados com a qualidade do ar inalado, possam contribuir para o seu agravamento. Para além dos aeroalergenos, outros factores ambientais como a poluição atmosférica estarão associados às doenças respiratórias. O ar respirado contém uma variedade de poluentes atmosféricos, provenientes quer de fontes naturais quer de origem antropogénica, nomeadamente de actividades industriais, domésticas ou das emissões de veículos. Estes poluentes, tradicionalmente considerados como um problema de foro ambiental, têm sido cada vez mais encarados como um problema de saúde pública. Também a qualidade do ar interior, tem sido associada a queixas respiratórias, não só em termos ocupacionais mas também em exposições domésticas. Dentro dos principais poluentes, encontramos a matéria particulada (como as PM10), o O3, NO2, e os compostos orgânicos voláteis (COVs). Se é verdade que os três primeiros têm como principais fontes de exposição a combustão fóssil associada aos veículos automóveis, já os COVs (como o benzeno, tolueno, xileno, etilbenzeno e formaldeído) são poluentes mais característicos do ar interior. Os mecanismos fisiopatológicos subjacentes à agressão dos poluentes do ar não se encontram convenientemente esclarecidos. Pensa-se que após a sua inalação, induzam um grau crescente de stress oxidativo que será responsável pelo desenvolvimento da inflamação das vias aéreas. A progressão do stress oxidativo e da inflamação, associarse- ão posteriormente a lesão local (pulmonar) e sistémica. Neste trabalho pretendeu-se avaliar os efeitos da exposição individual a diversos poluentes, do ar exterior e interior, sobre as vias aéreas, recorrendo a parâmetros funcionais, inflamatórios e do estudo do stress oxidativo. Neste sentido, desenvolveu-se um estudo de painel na cidade de Viseu, em que foram acompanhadas durante 18 meses, 51 crianças com história de sibilância, identificadas pelo questionário do estudo ISAAC. As crianças foram avaliadas em quatro Visitas (quatro medidas repetidas), através de diversos exames, que incluíram execução de espirometria com broncodilatação, medição ambulatória do PEF, medição de FENO e estudo do pH no condensado brônquico do ar exalado. O estudo dos 8-isoprostanos no condensado brônquico foi efectuado somente em duas Visitas, e o do doseamento de malonaldeído urinário somente na última Visita. Para além da avaliação do grau de infestação de ácaros do pó do colchão, para cada criança foi calculado o valor de exposição individual a PM10, O3, NO2, benzeno, tolueno, xileno, etilbenzeno e formaldeído, através de uma complexa metodologia que envolveu técnicas de modelação associadas a medições directas do ar interior (na casa e escola da criança) e do ar exterior. Para a análise de dados foram utilizadas equações de estimação generalizadas com uma matriz de correlação de trabalho uniforme, com excepção do estudo das associações entre poluentes, 8-isoprostanos e malonaldeído. Verificou-se na análise multivariável a existência de uma associação entre o agravamento dos parâmetros espirométricos e a exposição aumentada a PM10, NO2, benzeno, tolueno e etilbenzeno. Foram também encontradas associações entre diminuição do pH do EBC e exposição crescente a PM10, NO2, benzeno e etilbenzeno e associações entre valores aumentados de FENO e exposição a etilbenzeno e tolueno. O benzeno, o tolueno e o etilbenzeno foram associados com maior recurso a broncodilatador nos 6 meses anteriores à Visita e o tolueno com deslocações ao serviço de urgência. Os resultados dos modelos de regressão que incluíram o efeito do poluente ajustado para o grau de infestação de ácaros do pó foram, de uma forma geral, idênticos ao da análise multivariável anterior, com excepção das associações para com o FENO. Nos modelos de exposição com dois poluentes, com o FEV1 como variável resposta, somente o benzeno persistiu com significado estatístico. No modelo com dois poluentes tendo o pH do EBC como variável resposta, somente persistiram as PM10. Os 8-isoprostanos correlacionaram-se com alguns COVs, designadamente etilbenzeno, xileno, tolueno e benzeno. Os valores de malonaldeído urinário não se correlacionaram com os valores de poluentes. Verificou-se no entanto que de uma forma geral, e em particular mais uma vez para os COVs, as crianças mais expostas a poluentes, apresentaram valores superiores de malonaldeído na urina. Verificou-se que os poluentes do ar em geral, e os COVs em particular, se associaram com uma deterioração das vias aéreas. A exposição crescente a poluentes associou-se não só com obstrução brônquica, mas também com FENO aumentado e maior acidez das vias aéreas. A exposição crescente a COVs correlacionou-se com um maior stress oxidativo das vias aéreas (medido pelos 8-isoprostanos). As crianças com exposição superior a COVs apresentaram maiores valores de malonaldeído urinário. Este trabalho sugere uma associação entre exposição a poluentes, inflamação das vias aéreas e stress oxidativo. Vem reforçar o interesse dos poluentes do ar, nomeadamente os associados a ambientes interiores, frequentemente esquecidos e que poderão ser explicativos do agravamento duma criança com sibilância.-----------ABSTRACT: Asthma is a chronic respiratory disease that could be influenced by environmental factors, as allergens and air pollutants. The air breathed contains a diversity of air pollutants, both from natural or anthropogenic sources. Atmospheric pollution, traditionally considered an environmental problem, is nowadays looked as an important public health problem. Indoor air pollutants are also related with respiratory diseases, not only in terms of occupational exposures but also in domestic activities. Particulate matter (such as PM10), O3, NO2 and volatile organic compounds (VOCs) are the main air pollutants. The main source for PM10, O3, NO2 exposure is traffic exhaust while for VOCs (such as benzene, toluene, xylene, ethylbenzene and phormaldehyde) the main sources for exposure are located in indoor environments. The pathophysiologic mechanisms underlying the aggression of air pollutants are not properly understood. It is thought that after inhalation, air pollutants could induce oxidative stress, which would be responsible for airways inflammation. The progression of oxidative stress and airways inflammation, would contribute for the local and systemic effects of the air pollutants. The present study aimed to evaluate the effects of individual exposure to various pollutants over the airways, through lung function tests, inflammatory and oxidative stress biomarkers. In this sense, we developed a panel study in the city of Viseu, that included 51 children with a history of wheezing. Those children that were identified by the ISAAC questionnaire, were followed for 18 months. Children were assessed four times (four repeated measures) through the following tests: spirometry with bronchodilation test, PEF study, FENO evaluation and exhaled breath condensate pH measurement. 8-isoprostane in the exhaled breath condensate were also measured but only in two visits. Urinary malonaldehyde measurement was performed only in the last visit. Besides the assessment of the house dust mite infestation, we calculated for each child the value of individual exposure to a wide range of pollutants: PM10, O3, NO2, benzene, toluene, xylene, ethyl benzene and formaldehyde. This strategy used a complex methodology that included air pollution modelling techniques and direct measurements indoors (homes and schools) and outdoors. Generalized estimating equations with an exchangeable working correlation matrix were used for the analysis of the data. Exceptions were for the study of associations between air pollutants, malonaldehyde and 8-isoprostanes. In the multivariate analysis we found an association between worsening of spirometric outcomes and increased exposure to PM10, NO2, benzene, toluene and ethylbenzene. In the multivariate analysis we found also negative associations between EBC pH and exposure to PM10, NO2, benzene, ethylbenzene and positive associations between FENO and exposure to ethylbenzene and toluene. Benzene, toluene and ethylbenzene were associated with increased use of bronchodilator in the 6 months prior to the visit and toluene with emergency department visits. Results of the regression models that included also the effect of the pollutant adjusted for the degree of infestation to house dust mites, were identical to the previous models. Exceptions were for FENO associations. In the two-pollutant models, with the FEV1 as dependent variable, only benzene persisted with statistical significance. In the two pollutant model with pH of EBC as dependent variable, only PM10 persisted. 8-isoprostanes were well correlated with some VOCs, namely with ethylbenzene, xylene, toluene and benzene. Urinary malonaldehyde did not present any correlation with air pollutants exposure. However, those children more exposed to air pollutants (namely to VOCs), presented higher values of malonaldehyde. It was found that air pollutants in general, and namely VOCs, were associated with deterioration of the airways. The increased exposure to air pollutants was associated not only with airways obstruction, but also with increased FENO and higher acidity of the airways. The increased exposure to VOCs was correlated with increased airways oxidative stress (measured by 8-isoprostane). Children with higher levels of exposure to VOCs had higher values of urinary malonaldehyde. This study suggests a relation between air pollution, airways inflammation and oxidative stress. It suggests also that attention should be dedicated to air quality as air pollutants could cause airways deterioration.

Algorithm for the Parkinson’s disease behavioural models characterization using a biosensor

Dissertação para obtenção do Grau de Mestre em Engenharia Biomédica

DISCORDANCE BETWEEN BODY MASS INDEX AND ANTHROPOMETRIC MEASUREMENTS AMONG HIV-1-INFECTED PATIENTS ON ANTIRETROVIRAL THERAPY AND WITH LIPOATROPHY/LIPOHYPERTROPHY SYNDROME

Introduction: Highly Active Antiretroviral Therapy (HAART) has improved and extended the lives of thousands of people living with HIV/AIDS around the world. However, this treatment can lead to the development of adverse reactions such as lipoatrophy/lipohypertrophy syndrome (LLS) and its associated risks. Objective: This study was designed to assess the prevalence of self-reported lipodystrophy and nutritional status by anthropometric measurements in patients with HIV/AIDS. Methods: An observational study of 227 adult patients in the Secondary Immunodeficiencies Outpatient Department of Dermatology, Hospital das Clínicas, Faculty of Medicine, University of São Paulo (3002 ADEE-HCFMUSP). The sample was divided into three groups; Group 1 = 92 patients on HAART and with self-reported lipodystrophy, Group 2 = 70 patients on HAART without self-reported lipodystrophy and Group 3 = 65 patients not taking HAART. The nutritional status of individuals in the study sample was determined by body mass index (BMI) and percentage of body fat (% BF). The cardiovascular risk and diseases associated with abdominal obesity were determined by waist/hip ratio (WHR) and waist circumference (WC). Results: The prevalence of self-reported lipoatrophy/lipohypertrophy syndrome was 33% among women and 59% among men. Anthropometry showed depletion of fat mass in the evaluation of the triceps (TSF) in the treatment groups with HAART and was statistically independent of gender; for men p = 0.001, and for women p = 0.007. Similar results were found in the measurement of skin folds of the upper and lower body (p = 0.001 and p = 0.003 respectively). In assessing the nutritional status of groups by BMI and % BF, excess weight and body fat were more prevalent among women compared to men (p = 0.726). The WHR and WC revealed risks for cardiovascular and other diseases associated with abdominal obesity for women on HAART and with self-reported LLS (p = 0.005) and (p = 0.011). Conclusions: Anthropometric measurements were useful in the confirmation of the prevalence of LLS. BMI alone does not appear to be a good parameter for assessing the nutritional status of HIV-infected patients on HAART and with LLS. Other anthropometric measurements are needed to evaluate patients with the lipoatrophy/lipohypertrophy syndrome.

Development of novel human cellular models for neurotoxicity studies

Dissertation to obtain master degree in Genética Molecular e Biomedicina

Familial Abdominal Aortic Aneurysm Is Associated With More Complications After Endovascular Aneurysm Repair

OBJECTIVE: A familial predisposition to abdominal aortic aneurysms (AAAs) is present in approximately one-fifth of patients. Nevertheless, the clinical implications of a positive family history are not known. We investigated the risk of aneurysm-related complications after endovascular aneurysm repair (EVAR) for patients with and without a positive family history of AAA. METHODS: Patients treated with EVAR for intact AAAs in the Erasmus University Medical Center between 2000 and 2012 were included in the study. Family history was obtained by written questionnaire. Familial AAA (fAAA) was defined as patients having at least one first-degree relative affected with aortic aneurysm. The remaining patients were considered sporadic AAA. Cardiovascular risk factors, aneurysm morphology (aneurysm neck, aneurysm sac, and iliac measurements), and follow-up were obtained prospectively. The primary end point was complications after EVAR, a composite of endoleaks, need for secondary interventions, aneurysm sac growth, acute limb ischemia, and postimplantation rupture. Secondary end points were specific components of the primary end point (presence of endoleak, need for secondary intervention, and aneurysm sac growth), aneurysm neck growth, and overall survival. Kaplan-Meier estimates for the primary end point were calculated and compared using log-rank (Mantel-Cox) test of equality. A Cox-regression model was used to calculate the independent risk of complications associated with fAAA. RESULTS: A total of 255 patients were included in the study (88.6% men; age 72 ± 7 years, median follow-up 3.3 years; interquartile range, 2.2-6.1). A total of 51 patients (20.0%) were classified as fAAA. Patients with fAAA were younger (69 vs 72 years; P = .015) and were less likely to have ever smoked (58.8% vs 73.5%; P = .039). Preoperative aneurysm morphology was similar in both groups. Patients with fAAA had significantly more complications after EVAR (35.3% vs 19.1%; P = .013), with a twofold increased risk (adjusted hazard ratio, 2.1; 95% confidence interval, 1.2-3.7). Secondary interventions (39.2% vs 20.1%; P = .004) and aneurysm sac growth (20.8% vs 9.5%; P = .030) were the most important elements accounting for the difference. Furthermore, a trend toward more type I endoleaks during follow-up was observed (15.6% vs 7.4%; P = .063) and no difference in overall survival. CONCLUSIONS: The current study shows that patients with a familial form of AAA develop more aneurysm-related complications after EVAR, despite similar AAA morphology at baseline. These findings suggest that patients with fAAA form a specific subpopulation and create awareness for a possible increase in the risk of complications after EVAR.

Invasive Assessment of the Coronary Microcirculation Using the Index of Microcirculatory Resistance: Description and Validation of an Animal Model

INTRODUCTION: The index of microcirculatory resistance (IMR) enables/provides quantitative, invasive, and real-time assessment of coronary microcirculation status. AIMS: The primary aim of this study was to validate the assessment of IMR in a large animal model, and the secondary aim was to compare two doses of intracoronary papaverine, 5 and 10 mg, for induction of maximal hyperemia and its evolution over time. METHODS: Measurements of IMR were performed in eight pigs. Mean distal pressure (Pd) and mean transit time (Tmn) were measured at rest and at maximal hyperemia induced with intracoronary papaverine, 5 and 10 mg, and after 2, 5, 8 and 10 minutes. Disruption of the microcirculation was achieved by selective injection of 40-μm microspheres via a microcatheter in the left anterior descending artery. RESULTS: In each animal 14 IMR measurements were made. There were no differences between the two doses of papaverine regarding Pd response and IMR values - 11 ± 4.5 U with 5 mg and 10.6 ± 3 U with 10 mg (p=0.612). The evolution of IMR over time was also similar with the two doses, with significant differences from resting values disappearing after five minutes of intracoronary papaverine administration. IMR increased with disrupted microcirculation in all animals (41 ± 16 U, p=0.001). CONCLUSIONS: IMR provides invasive and real-time assessment of coronary microcirculation. Disruption of the microvascular bed is associated with a significant increase in IMR. A 5-mg dose of intracoronary papaverine is as effective as a 10-mg dose in inducing maximal hyperemia. After five minutes of papaverine administration there is no significant difference from resting hemodynamic status.