Markov blanket-based approach for learning multi-dimensional Bayesian network classifiers: An application to predict the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson's Disease Questionnaire (PDQ-39)

Multi-dimensional Bayesian network classifiers (MBCs) are probabilistic graphical models recently proposed to deal with multi-dimensional classification problems, where each instance in the data set has to be assigned to more than one class variable. In this paper, we propose a Markov blanket-based approach for learning MBCs from data. Basically, it consists of determining the Markov blanket around each class variable using the HITON algorithm, then specifying the directionality over the MBC subgraphs. Our approach is applied to the prediction problem of the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson’s Disease Questionnaire (PDQ-39) in order to estimate the health-related quality of life of Parkinson’s patients. Fivefold cross-validation experiments were carried out on randomly generated synthetic data sets, Yeast data set, as well as on a real-world Parkinson’s disease data set containing 488 patients. The experimental study, including comparison with additional Bayesian network-based approaches, back propagation for multi-label learning, multi-label k-nearest neighbor, multinomial logistic regression, ordinary least squares, and censored least absolute deviations, shows encouraging results in terms of predictive accuracy as well as the identification of dependence relationships among class and feature variables.

Socio-demographic factors, smoking, symptoms, morbidities and pulmonary function and quality of life in individuals with a heavy smoking history

Objective: To determine which socio-demographic, exposure, morbidity and symptom variables are associated with health-related quality of life among former and current heavy smokers. Methods: Cross sectional data from 2537 participants were studied. All participants were at ≥2% risk of developing lung cancer within 6 years. Linear and logistic regression models utilizing a multivariable fractional polynomial selection process identified variables associated with health-related quality of life, measured by the EQ-5D. Results: Upstream and downstream associations between smoking cessation and higher health-related quality of life were evident. Significant upstream associations, such as education level and current working status and were explained by the addition of morbidities and symptoms to regression models. Having arthritis, decreased forced expiratory volume in one second, fatigue, poor appetite or dyspnea were most highly and commonly associated with decreased HRQoL. Discussion: Upstream factors such as educational attainment, employment status and smoking cessation should be targeted to prevent decreased health-related quality of life. Practitioners should focus treatment on downstream factors, especially symptoms, to improve health-related quality of life.

Not all roads lead to Rome - a review of quality of life measurement in diabetes

Aims Quality of life (QoL) is recognized widely as an important health outcome in diabetes, where the burden of selfmanagement places great demands on the individual. However, the concept of QoL remains ambiguous and poorly defined. The aim of our review is to clarify the measurement of QoL in terms of conceptualization, terminology and psychometric properties, to review the instruments that have been used most frequently to assess QoL in diabetes research and make recommendations for how to select measures appropriately.

Methods A systematic literature search was conducted to identify the ten measures most frequently used to assess QoL in diabetes research (including clinical trials) from 1995 to March 2008.

Results Six thousand and eight-five abstracts were identified and screened for instrument names. Of the ten instruments most frequently used to assess ‘QoL’, only three actually do so [i.e. the generic World Health Organization Quality of Life (WHOQOL) and the diabetes-specific Diabetes Quality of Life (DQOL) and Audit of Diabetes-Dependent Quality of Life (ADDQoL)]. Seven instruments more accurately measure health status [Short-Form 36 (SF-36), EuroQoL 5-Dimension (EQ-5D)], treatment satisfaction [Diabetes Treatment Satisfaction Questionnaire (DTSQ)] and psychological well-being [Beck Depression Inventory (BDI), Hospital Anxiety and Depression Scale (HADS), Well-Being Questionnaire (W-BQ), Problem Areas in Diabetes (PAID)].

Conclusions No single measure can suit every purpose or application but, when measures are selected inappropriately and data misinterpreted, any conclusions drawn are fundamentally flawed. If we value QoL as a therapeutic goal, we must ensure that the instruments we use are both valid and reliable. QoL assessment has the proven potential to identify ways in which treatments can be tailored to reduce the burden of diabetes. With careful consideration, appropriate measures can be selected and truly robust assessments undertaken successfully.

Predicting EQ-5D from the Parkinson's disease questionnaire PDQ-8 using multi-dimensional Bayesian network classifiers

The impact of the Parkinson's disease and its treatment on the patients' health-related quality of life can be estimated either by means of generic measures such as the european quality of Life-5 Dimensions (EQ-5D) or specific measures such as the 8-item Parkinson's disease questionnaire (PDQ-8). In clinical studies, PDQ-8 could be used in detriment of EQ-5D due to the lack of resources, time or clinical interest in generic measures. Nevertheless, PDQ-8 cannot be applied in cost-effectiveness analyses which require generic measures and quantitative utility scores, such as EQ-5D. To deal with this problem, a commonly used solution is the prediction of EQ-5D from PDQ-8. In this paper, we propose a new probabilistic method to predict EQ-5D from PDQ-8 using multi-dimensional Bayesian network classifiers. Our approach is evaluated using five-fold cross-validation experiments carried out on a Parkinson's data set containing 488 patients, and is compared with two additional Bayesian network-based approaches, two commonly used mapping methods namely, ordinary least squares and censored least absolute deviations, and a deterministic model. Experimental results are promising in terms of predictive performance as well as the identification of dependence relationships among EQ-5D and PDQ-8 items that the mapping approaches are unable to detect

Physical activity and health related quality of life

Background Research on the relationship between Health Related Quality of Life (HRQoL) and physical activity (PA), to date, have rarely investigated how this relationship differ across objective and subjective measures of PA. The aim of this paper is to explore the relationship between HRQoL and PA, and examine how this relationship differs across objective and subjective measures of PA, within the context of a large representative national survey from England. Methods Using a sample of 5,537 adults (40–60 years) from a representative national survey in England (Health Survey for England 2008), Tobit regressions with upper censoring was employed to model the association between HRQoL and objective, and subjective measures of PA controlling for potential confounders. We tested the robustness of this relationship across specific types of PA. HRQoL was assessed using the summary measure of health state utility value derived from the EuroQol-5 Dimensions (EQ-5D) whilst PA was assessed via subjective measure (questionnaire) and objective measure (accelerometer- actigraph model GT1M). The actigraph was worn (at the waist) for 7 days (during waking hours) by a randomly selected sub-sample of the HSE 2008 respondents (4,507 adults – 16 plus years), with a valid day constituting 10 hours. Analysis was conducted in 2010. Results Findings suggest that higher levels of PA are associated with better HRQoL (regression coefficient: 0.026 to 0.072). This relationship is consistent across different measures and types of PA although differences in the magnitude of HRQoL benefit associated with objective and subjective (regression coefficient: 0.047) measures of PA are noticeable, with the former measure being associated with a relatively better HRQoL (regression coefficient: 0.072). Conclusion Higher levels of PA are associated with better HRQoL. Using an objective measure of PA compared with subjective shows a relatively better HRQoL.

Social/economic costs and health-related quality of life in patients with epidermolysis bullosa in Europe

BACKGROUND: The aim of this study was to determine the social/economic costs and health-related quality of life (HRQOL) of patients with epidermolysis bullosa (EB) in eight EU member states. METHODS: We conducted a cross-sectional study of patients with EB from Bulgaria, France, Germany, Hungary, Italy, Spain, Sweden and the United Kingdom. Data on demographic characteristics, health resource utilisation, informal care, labour productivity losses, and HRQOL were collected from the questionnaires completed by patients or their caregivers. HRQOL was measured with the EuroQol 5-domain (EQ-5D) questionnaire. RESULTS: A total of 204 patients completed the questionnaire. Average annual costs varied from country to country, and ranged from euro9509 to euro49,233 (reference year 2012). Estimated direct healthcare costs ranged from euro419 to euro10,688; direct non-healthcare costs ranged from euro7449 to euro37,451 and labour productivity losses ranged from euro0 to euro7259. The average annual cost per patient across all countries was estimated at euro31,390, out of which euro5646 accounted for direct health costs (18.0 %), euro23,483 accounted for direct non-healthcare costs (74.8 %), and euro2261 accounted for indirect costs (7.2 %). Costs were shown to vary across patients with different disability but also between children and adults. The mean EQ-5D score for adult EB patients was estimated at between 0.49 and 0.71 and the mean EQ-5D visual analogue scale score was estimated at between 62 and 77. CONCLUSION: In addition to its negative impact on patient HRQOL, our study indicates the substantial social/economic burden of EB in Europe, attributable mostly to high direct non-healthcare costs.

Telephone reliability of the Frenchay Activity Index and EQ-5D amongst older adults

Background Older adults may find it problematic to attend hospital appointments due to the difficulty associated with travelling to, within and from a hospital facility for the purpose of a face-to-face assessment. This study aims to investigate equivalence between telephone and face-to-face administration for the Frenchay Activities Index (FAI) and the Euroqol-5D (EQ-5D) generic health-related quality of life instrument amongst an older adult population. Methods Patients aged >65 (n = 53) who had been discharged to the community following an acute hospital admission underwent telephone administration of the FAI and EQ-5D instruments seven days prior to attending a hospital outpatient appointment where they completed a face-to-face administration of these instruments. Results Overall, 40 subjects' datasets were complete for both assessments and included in analysis. The FAI items had high levels of agreement between the two modes of administration (item kappa's ranged 0.73 to 1.00) as did the EQ-5D (item kappa's ranged 0.67–0.83). For the FAI, EQ-5D VAS and EQ-5D utility score, intraclass correlation coefficients were 0.94, 0.58 and 0.82 respectively with paired t-tests indicating no significant systematic difference (p = 0.100, p = 0.690 and p = 0.290 respectively). Conclusion Telephone administration of the FAI and EQ-5D instruments provides comparable results to face-to-face administration amongst older adults deemed to have cognitive functioning intact at a basic level, indicating that this is a suitable alternate approach for collection of this information.

Assessment of Health-Related Quality of Life using the EQ-5D-3L in Individuals with Type 2 Diabetes Mellitus

The daily experience with type 2 diabetes mellitus (T2DM) has significant adverse effects on health-related quality of life (HRQoL). HRQoL assessment is essential for measuring the impact of the disease on the patient and selecting individualized strategies. Generic measures for assessing HRQoL are very useful because, unlike specific measurement instruments, they allow for the comparison with other instruments. The EQ-5D-3L is a generic measure and it describes HRQoL in five dimensions; mobility, self-care, usual activities, pain/discomfort and anxiety/depression, with three levels each. In Portugal, studies using the EQ-5D-3L as a generic measure to assess HRQoL in diabetic patients are scarce. Objective: To assess HRQoL in individuals with T2DM using the Portuguese version of the EQ-5D-3L. Methodology: An accidental sample of patients with T2DM (n=437) was selected at Family Health Units and healthcare centers in Coimbra, Portugal, between January 2013 and January 2014. The EQ-5D-3L was applied in interviews. The EQ-5D-3L score was calculated based on the answers to the five dimensions and the value system for the Portuguese population. Results: In this sample, 100% of the participants answered the EQ-5D-3L. The HRQoL score was 0.6772 in the EQ-5D-3L and 64.85 in the EQ-VAS. The most frequent answers to the five dimensions were no problems or some problems. The mean score of the EQ-5D-3L was significantly associated with age, male gender, high level of education, having an occupation, practicing physical activity, being single and having been diagnosed with T2DM for less time. The Cronbach alpha's value was 0.674, confirming an acceptable internal consistency. Conclusion: HRQoL levels in individuals with T2DM are lower than the national average and vary depending on sociodemographic and clinical characteristics. The EQ-5D-3L is a reliable instrument that can be used to assess the quality of life of diabetic patients and contribute to assess the patients' overall health status, adding data from the subjective dimension of self-care management.

Cross cultural adaptation and validation of a Spanish version of the lower limb functional index

Background: The Lower Limb Functional Index (LLFI) is a relatively recently published regional outcome measure. The development article showed the LLFI had robust and valid clinimetric properties with sound psychometric and practical characteristics when compared to the Lower Limb Extremity Scale (LEFS) criterion standard. Objective: The purpose of this study was cross cultural adaptation and validation of the LLFI Spanish-version (LLFI-Sp) in a Spanish population. Methods: A two stage observational study was conducted. The LLFI was initially cross-culturally adapted to Spanish through double forward and single backward translation; then subsequently validated for the psychometric characteristics of validity, internal consistency, reliability, error score and factor structure. Participants (n = 136) with various lower limb conditions of >12 weeks duration completed the LLFI-Sp, Western Ontario and McMaster University Osteoarthritis Index (WOMAC) and the Euroqol Health Questionnaire 5 Dimensions (EQ-5D-3 L). The full sample was employed to determine internal consistency, concurrent criterion validity, construct validity and factor structure; a subgroup (n = 45) determined reliability at seven days concurrently completing a global rating of change scale. Results: The LLFI-Sp demonstrated high but not excessive internal consistency (α = 0.91) and high reliability (ICC = 0.96). The factor structure was one-dimensional which supported the construct validity. Criterion validity with the WOMAC was strong (r = 0.77) and with the EQ-5D-3 L fair and inversely correlated (r = -0.62). The study limitations included the lack of longitudinal data and the determination of responsiveness. Conclusions: The LLFI-Sp supports the findings of the original English version as being a valid lower limb regional outcome measure. It demonstrated similar psychometric properties for internal consistency, validity, reliability, error score and factor structure.

Cross-cultural adaptation, reliability and validity of the Spanish version of the upper limb functional index

Background The Upper Limb Functional Index (ULFI) is an internationally widely used outcome measure with robust, valid psychometric properties. The purpose of study is to develop and validate a ULFI Spanish-version (ULFI-Sp). Methods A two stage observational study was conducted. The ULFI was cross-culturally adapted to Spanish through double forward and backward translations, the psychometric properties were then validated. Participants (n = 126) with various upper limb conditions of >12 weeks duration completed the ULFI-Sp, QuickDASH and the Euroqol Health Questionnaire 5 Dimensions (EQ-5D-3 L). The full sample determined internal consistency, concurrent criterion validity, construct validity and factor structure; a subgroup (n = 35) determined reliability at seven days. Results The ULFI-Sp demonstrated high internal consistency (α = 0.94) and reliability (r = 0.93). Factor structure was one-dimensional and supported construct validity. Criterion validity with the EQ-5D-3 L was fair and inversely correlated (r = −0.59). The QuickDASH data was unavailable for analysis due to excessive missing responses. Conclusions The ULFI-Sp is a valid upper limb outcome measure with similar psychometric properties to the English language version.

Comparing the performance of the EQ-5D-3L and the EQ-5D-5L in young Portuguese adults

Background Some studies have reported a ceiling effect in EQ-5D-3L, especially in healthy and/or young individuals. Recently, two further levels have been included in its measurement model (EQ-5D-5L). The purposes of this study were (1) to assess the properties of the EQ-5D-5L in comparison with the standard EQ-5D-3L in a sample of young adults, (2) to foreground the importance of collecting qualitative data to confirm, validate or refine the EQ-5D questionnaire items and (3) to raise questions pertaining to the wording in these questionnaire items. Methods The data used came from a sample of respondents aged 30 or under (n = 624). They completed both versions of the EQ-5D, which were compared in terms of feasibility, level of inconsistency and ceiling effect. Agreement between the instruments was assessed using correlation coefficients and Bland-Altman plots. Known-groups validity of the EQ-5D-5L was also assessed using non-parametric tests. The discriminative properties were compared using receiver operating characteristic curves. Finally, four interviews were conducted for retrospective reports to elicit respondents’ understanding and perceptions of the format, instructions, items, and responses. Results Quantitative results show a ceiling effect reduction of 25.3 % and a high level agreement between both indices. Known-groups validity was confirmed for the EQ-5D-5L. Explorative interviews indicated ambiguity and low degree of certainty in regards to conceptualizing differences between levels moderate-slight across three dimensions. Conclusions The EQ-5D-5L performed better than the EQ-5D-3L. However, the explorative interviews demonstrated several limitations in the EQ-5D questionnaire wording and high context-dependent answers point to lack of illnesses’ experience amongst young adults.

基于广义正交多项式迭积微分算子的地震波场数值模拟研究

Seismic Numerical Modeling is one of bases of the Exploratory Seismology and Academic Seismology, also is a research field in great demand. Essence of seismic numerical modeling is to assume that structure and parameters of the underground media model are known, simulate the wave-field and calculate the numerical seismic record that should be observed. Seismic numerical modeling is not only a means to know the seismic wave-field in complex inhomogeneous media, but also a test to the application effect by all kinds of methods. There are many seismic numerical modeling methods, each method has its own merits and drawbacks. During the forward modeling, the computation precision and the efficiency are two pivotal questions to evaluate the validity and superiority of the method. The target of my dissertation is to find a new method to possibly improve the computation precision and efficiency, and apply the new forward method to modeling the wave-field in the complex inhomogeneous media. Convolutional Forsyte polynomial differentiator (CFPD) approach developed in this dissertation is robust and efficient, it shares some of the advantages of the high precision of generalized orthogonal polynomial and the high speed of the short operator finite-difference. By adjusting the operator length and optimizing the operator coefficient, the method can involve whole and local information of the wave-field. One of main tasks of the dissertation is to develop a creative, generalized and high precision method. The author introduce convolutional Forsyte polynomial differentiator to calculate the spatial derivative of seismic wave equation, and apply the time staggered grid finite-difference which can better meet the high precision of the convolutional differentiator to substitute the conventional finite-difference to calculate the time derivative of seismic wave equation, then creating a new forward method to modeling the wave-field in complex inhomogeneous media. Comparing with Fourier pseudo-spectral method, Chebyshev pseudo-spectral method, staggered- grid finite difference method and finite element method, convolutional Forsyte polynomial differentiator (CFPD) method has many advantages: 1. Comparing with Fourier pseudo-spectral method. Fourier pseudo-spectral method (FPS) is a local operator, its results have Gibbs effects when the media parameters change, then arose great errors. Therefore, Fourier pseudo-spectral method can not deal with special complex and random heterogeneous media. But convolutional Forsyte polynomial differentiator method can cover global and local information. So for complex inhomogeneous media, CFPD is more efficient. 2. Comparing with staggered-grid high-order finite-difference method, CFPD takes less dots than FD at single wave length, and the number does not increase with the widening of the studying area. 3. Comparing with Chebyshev pseudo-spectral method (CPS). The calculation region of Chebyshev pseudo-spectral method is fixed in , under the condition of unchangeable precision, the augmentation of calculation is unacceptable. Thus Chebyshev pseudo-spectral method is inapplicable to large area. CFPD method is more applicable to large area. 4. Comparing with finite element method (FE), CFPD can use lager grids. The other task of this dissertation is to study 2.5 dimension (2.5D) seismic wave-field. The author reviews the development and present situation of 2.5D problem, expatiates the essentiality of studying the 2.5D problem, apply CFPD method to simulate the seismic wave-field in 2.5D inhomogeneous media. The results indicate that 2.5D numerical modeling is efficient to simulate one of the sections of 3D media, 2.5D calculation is much less time-consuming than 3D calculation, and the wave dispersion of 2.5D modeling is obviously less than that of 3D modeling. Question on applying time staggered-grid convolutional differentiator based on CFPD to modeling 2.5D complex inhomogeneous media was not studied by any geophysicists before, it is a fire-new creation absolutely. The theory and practices prove that the new method can efficiently model the seismic wave-field in complex media. Proposing and developing this new method can provide more choices to study the seismic wave-field modeling, seismic wave migration, seismic inversion, and seismic wave imaging.

Multi-dimensional classification using Bayesian networks for stationary and evolving streaming data

Hoy en día, con la evolución continua y rápida de las tecnologías de la información y los dispositivos de computación, se recogen y almacenan continuamente grandes volúmenes de datos en distintos dominios y a través de diversas aplicaciones del mundo real. La extracción de conocimiento útil de una cantidad tan enorme de datos no se puede realizar habitualmente de forma manual, y requiere el uso de técnicas adecuadas de aprendizaje automático y de minería de datos. La clasificación es una de las técnicas más importantes que ha sido aplicada con éxito a varias áreas. En general, la clasificación se compone de dos pasos principales: en primer lugar, aprender un modelo de clasificación o clasificador a partir de un conjunto de datos de entrenamiento, y en segundo lugar, clasificar las nuevas instancias de datos utilizando el clasificador aprendido. La clasificación es supervisada cuando todas las etiquetas están presentes en los datos de entrenamiento (es decir, datos completamente etiquetados), semi-supervisada cuando sólo algunas etiquetas son conocidas (es decir, datos parcialmente etiquetados), y no supervisada cuando todas las etiquetas están ausentes en los datos de entrenamiento (es decir, datos no etiquetados). Además, aparte de esta taxonomía, el problema de clasificación se puede categorizar en unidimensional o multidimensional en función del número de variables clase, una o más, respectivamente; o también puede ser categorizado en estacionario o cambiante con el tiempo en función de las características de los datos y de la tasa de cambio subyacente. A lo largo de esta tesis, tratamos el problema de clasificación desde tres perspectivas diferentes, a saber, clasificación supervisada multidimensional estacionaria, clasificación semisupervisada unidimensional cambiante con el tiempo, y clasificación supervisada multidimensional cambiante con el tiempo. Para llevar a cabo esta tarea, hemos usado básicamente los clasificadores Bayesianos como modelos. La primera contribución, dirigiéndose al problema de clasificación supervisada multidimensional estacionaria, se compone de dos nuevos métodos de aprendizaje de clasificadores Bayesianos multidimensionales a partir de datos estacionarios. Los métodos se proponen desde dos puntos de vista diferentes. El primer método, denominado CB-MBC, se basa en una estrategia de envoltura de selección de variables que es voraz y hacia delante, mientras que el segundo, denominado MB-MBC, es una estrategia de filtrado de variables con una aproximación basada en restricciones y en el manto de Markov. Ambos métodos han sido aplicados a dos problemas reales importantes, a saber, la predicción de los inhibidores de la transcriptasa inversa y de la proteasa para el problema de infección por el virus de la inmunodeficiencia humana tipo 1 (HIV-1), y la predicción del European Quality of Life-5 Dimensions (EQ-5D) a partir de los cuestionarios de la enfermedad de Parkinson con 39 ítems (PDQ-39). El estudio experimental incluye comparaciones de CB-MBC y MB-MBC con los métodos del estado del arte de la clasificación multidimensional, así como con métodos comúnmente utilizados para resolver el problema de predicción de la enfermedad de Parkinson, a saber, la regresión logística multinomial, mínimos cuadrados ordinarios, y mínimas desviaciones absolutas censuradas. En ambas aplicaciones, los resultados han sido prometedores con respecto a la precisión de la clasificación, así como en relación al análisis de las estructuras gráficas que identifican interacciones conocidas y novedosas entre las variables. La segunda contribución, referida al problema de clasificación semi-supervisada unidimensional cambiante con el tiempo, consiste en un método nuevo (CPL-DS) para clasificar flujos de datos parcialmente etiquetados. Los flujos de datos difieren de los conjuntos de datos estacionarios en su proceso de generación muy rápido y en su aspecto de cambio de concepto. Es decir, los conceptos aprendidos y/o la distribución subyacente están probablemente cambiando y evolucionando en el tiempo, lo que hace que el modelo de clasificación actual sea obsoleto y deba ser actualizado. CPL-DS utiliza la divergencia de Kullback-Leibler y el método de bootstrapping para cuantificar y detectar tres tipos posibles de cambio: en las predictoras, en la a posteriori de la clase o en ambas. Después, si se detecta cualquier cambio, un nuevo modelo de clasificación se aprende usando el algoritmo EM; si no, el modelo de clasificación actual se mantiene sin modificaciones. CPL-DS es general, ya que puede ser aplicado a varios modelos de clasificación. Usando dos modelos diferentes, el clasificador naive Bayes y la regresión logística, CPL-DS se ha probado con flujos de datos sintéticos y también se ha aplicado al problema real de la detección de código malware, en el cual los nuevos ficheros recibidos deben ser continuamente clasificados en malware o goodware. Los resultados experimentales muestran que nuestro método es efectivo para la detección de diferentes tipos de cambio a partir de los flujos de datos parcialmente etiquetados y también tiene una buena precisión de la clasificación. Finalmente, la tercera contribución, sobre el problema de clasificación supervisada multidimensional cambiante con el tiempo, consiste en dos métodos adaptativos, a saber, Locally Adpative-MB-MBC (LA-MB-MBC) y Globally Adpative-MB-MBC (GA-MB-MBC). Ambos métodos monitorizan el cambio de concepto a lo largo del tiempo utilizando la log-verosimilitud media como métrica y el test de Page-Hinkley. Luego, si se detecta un cambio de concepto, LA-MB-MBC adapta el actual clasificador Bayesiano multidimensional localmente alrededor de cada nodo cambiado, mientras que GA-MB-MBC aprende un nuevo clasificador Bayesiano multidimensional. El estudio experimental realizado usando flujos de datos sintéticos multidimensionales indica los méritos de los métodos adaptativos propuestos. ABSTRACT Nowadays, with the ongoing and rapid evolution of information technology and computing devices, large volumes of data are continuously collected and stored in different domains and through various real-world applications. Extracting useful knowledge from such a huge amount of data usually cannot be performed manually, and requires the use of adequate machine learning and data mining techniques. Classification is one of the most important techniques that has been successfully applied to several areas. Roughly speaking, classification consists of two main steps: first, learn a classification model or classifier from an available training data, and secondly, classify the new incoming unseen data instances using the learned classifier. Classification is supervised when the whole class values are present in the training data (i.e., fully labeled data), semi-supervised when only some class values are known (i.e., partially labeled data), and unsupervised when the whole class values are missing in the training data (i.e., unlabeled data). In addition, besides this taxonomy, the classification problem can be categorized into uni-dimensional or multi-dimensional depending on the number of class variables, one or more, respectively; or can be also categorized into stationary or streaming depending on the characteristics of the data and the rate of change underlying it. Through this thesis, we deal with the classification problem under three different settings, namely, supervised multi-dimensional stationary classification, semi-supervised unidimensional streaming classification, and supervised multi-dimensional streaming classification. To accomplish this task, we basically used Bayesian network classifiers as models. The first contribution, addressing the supervised multi-dimensional stationary classification problem, consists of two new methods for learning multi-dimensional Bayesian network classifiers from stationary data. They are proposed from two different points of view. The first method, named CB-MBC, is based on a wrapper greedy forward selection approach, while the second one, named MB-MBC, is a filter constraint-based approach based on Markov blankets. Both methods are applied to two important real-world problems, namely, the prediction of the human immunodeficiency virus type 1 (HIV-1) reverse transcriptase and protease inhibitors, and the prediction of the European Quality of Life-5 Dimensions (EQ-5D) from 39-item Parkinson’s Disease Questionnaire (PDQ-39). The experimental study includes comparisons of CB-MBC and MB-MBC against state-of-the-art multi-dimensional classification methods, as well as against commonly used methods for solving the Parkinson’s disease prediction problem, namely, multinomial logistic regression, ordinary least squares, and censored least absolute deviations. For both considered case studies, results are promising in terms of classification accuracy as well as regarding the analysis of the learned MBC graphical structures identifying known and novel interactions among variables. The second contribution, addressing the semi-supervised uni-dimensional streaming classification problem, consists of a novel method (CPL-DS) for classifying partially labeled data streams. Data streams differ from the stationary data sets by their highly rapid generation process and their concept-drifting aspect. That is, the learned concepts and/or the underlying distribution are likely changing and evolving over time, which makes the current classification model out-of-date requiring to be updated. CPL-DS uses the Kullback-Leibler divergence and bootstrapping method to quantify and detect three possible kinds of drift: feature, conditional or dual. Then, if any occurs, a new classification model is learned using the expectation-maximization algorithm; otherwise, the current classification model is kept unchanged. CPL-DS is general as it can be applied to several classification models. Using two different models, namely, naive Bayes classifier and logistic regression, CPL-DS is tested with synthetic data streams and applied to the real-world problem of malware detection, where the new received files should be continuously classified into malware or goodware. Experimental results show that our approach is effective for detecting different kinds of drift from partially labeled data streams, as well as having a good classification performance. Finally, the third contribution, addressing the supervised multi-dimensional streaming classification problem, consists of two adaptive methods, namely, Locally Adaptive-MB-MBC (LA-MB-MBC) and Globally Adaptive-MB-MBC (GA-MB-MBC). Both methods monitor the concept drift over time using the average log-likelihood score and the Page-Hinkley test. Then, if a drift is detected, LA-MB-MBC adapts the current multi-dimensional Bayesian network classifier locally around each changed node, whereas GA-MB-MBC learns a new multi-dimensional Bayesian network classifier from scratch. Experimental study carried out using synthetic multi-dimensional data streams shows the merits of both proposed adaptive methods.