Search

[EN] The subjects Health Information Systems and Telemedicine and Data Quality and Interoperability of the Degree and Master in Biomedical Engineering of the Universitat Politècnica de València, Spain, address learning outcomes related to managing and processing biomedical databases, using health information standards for data capture and exchange, data quality assessment, and developing machine-learning models from these data. These learning outcomes cover a large range of distinct activities in the biomedical data life-cycle, what may hinder the learning process in the limited time assigned for the subject. We propose a project based learning approach addressing the full life-cycle of biomedical data on the MIMIC-III (Medical Information Mart for Intensive Care III) Open Dataset, a freely accessible database comprising information relating to patients admitted to critical care units. By means of this active learning approach, students can achieve all the learning outcomes of the subject in an integrated manner: understanding the MIMIC-III data model, using health information standards such as International Classification of Diseases 9th Edition (ICD-9), mapping to interoperability standards, querying data, creating data tables and addressing data quality towards applying reliable statistical and machine learning analysis and, developing predictive models for several tasks such as predicting in-hospital mortality. MIMIC-III is widely used in the academia and science, with a large amount of publicly available resources and scientific articles to support the students learning. Additionally, the students will gain new competences in the use of Open Data and Research Ethics and Compliance Training.

[ES] El presente Trabajo Fin de Grado (TFG) tiene como objetivo diseñar la interfaz gráfica de un Sistemas de Ayuda a la Decisión Médica (SADM) dedicado a la predicción de la evolución de pacientes con COVID-19. Para el diseño de la interfaz gráfica se tendrá en cuenta el contexto de uso del SADM, el perfil del/la usuario/a y las tareas a realizar con el sistema. El TFG está enmarcado en el proyecto "SUBCOVERWD-19" liderado por la UPV en la convocatoria FONDO SUPERA COVID-19 de la CRUE-Banco Santander. En el TFG se aplicarán técnicas de diseño de interfaces gráficas centradas en el usuario, se evaluará su usabilidad, así como la experiencia del usuario, todo ello con el objetivo de favorecer la aceptabilidad del sistema por los profesionales, y utilización en el ámbito médico., [EN] The objective of this Final Degree Project (TFG) is to design the graphical interface of a Medical Decision Aid Systems (SADM) dedicated to the prediction of the evolution of patients with COVID-19. For the design of the graphical interface, the context of use of the SADM, the profile of the user and the tasks to be carried out with the system will be taken into account. The TFG is framed in the project "SUBCOVERWD-19" led by the UPV in the call FONDO SUPERA COVID-19 of the CRUE-Banco Santander. In the TFG, user-centered graphic interface design techniques will be applied, their usability will be evaluated, as well as the user experience, all with the aim of favoring the acceptability of the system by professionals, and use in the medical field.

[EN] The objective of this work was to develop a predictive model to aid non-clinical dispatchers to classify emergency medical call incidents by their life-threatening level (yes/no), admissible response delay (undelayable, minutes, hours, days) and emergency system jurisdiction (emergency system/primary care) in real time. We used a total of 1 244 624 independent incidents from the Valencian emergency medical dispatch service in Spain, compiled in retrospective from 2009 to 2012, including clinical features, demographics, circumstantial factors and free text dispatcher observations. Based on them, we designed and developed DeepEMC2, a deep ensemble multitask model integrating four subnetworks: three specialized to context, clinical and text data, respectively, and another to ensemble the former. The four subnetworks are composed in turn by multi-layer perceptron modules, bidirectional long short-term memory units and a bidirectional encoding representations from transformers module. DeepEMC2 showed a macro F1-score of 0.759 in life-threatening classification, 0.576 in admissible response delay and 0.757 in emergency system jurisdiction. These results show a substantial performance increase of 12.5 %, 17.5 % and 5.1 %, respectively, with respect to the current in-house triage protocol of the Valencian emergency medical dispatch service. Besides, DeepEMC2 significantly outperformed a set of baseline machine learning models, including naive bayes, logistic regression, random forest and gradient boosting (¿ = 0.05). Hence, DeepEMC2 is able to: 1) capture information present in emergency medical calls not considered by the existing triage protocol, and 2) model complex data dependencies not feasible by the tested baseline models. Likewise, our results suggest that most of this unconsidered information is present in the free text dispatcher observations. To our knowledge, this study describes the first deep learning model undertaking emergency medical call incidents cla

This is the peer reviewed version of the following article: del Mar Álvarez-Torres, M., Juan-Albarracín, J., Fuster-Garcia, E., Bellvís-Bataller, F., Lorente, D., Reynés, G., Font de Mora, J., Aparici-Robles, F., Botella, C., Muñoz-Langa, J., Faubel, R., Asensio-Cuesta, S., García-Ferrando, G.A., Chelebian, E., Auger, C., Pineda, J., Rovira, A., Oleaga, L., Mollà-Olmos, E., Revert, A.J., Tshibanda, L., Crisi, G., Emblem, K.E., Martin, D., Due-Tønnessen, P., Meling, T.R., Filice, S., Sáez, C. and García-Gómez, J.M. (2020), Robust association between vascular habitats and patient prognosis in glioblastoma: An international multicenter study. J Magn Reson Imaging, 51: 1478-1486, which has been published in final form at https://doi.org/10.1002/jmri.26958. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Self-Archiving., [EN] Background Glioblastoma (GBM) is the most aggressive primary brain tumor, characterized by a heterogeneous and abnormal vascularity. Subtypes of vascular habitats within the tumor and edema can be distinguished: high angiogenic tumor (HAT), low angiogenic tumor (LAT), infiltrated peripheral edema (IPE), and vasogenic peripheral edema (VPE). Purpose To validate the association between hemodynamic markers from vascular habitats and overall survival (OS) in glioblastoma patients, considering the intercenter variability of acquisition protocols. Study Type Multicenter retrospective study. Population In all, 184 glioblastoma patients from seven European centers participating in the NCT03439332 clinical study. Field Strength/Sequence 1.5T (for 54 patients) or 3.0T (for 130 patients). Pregadolinium and postgadolinium-based contrast agent-enhanced T-1-weighted MRI, T-2- and FLAIR T-2-weighted, and dynamic susceptibility contrast (DSC) T-2* perfusion. Assessment We analyzed preoperative MRIs to establish the association between the maximum relative cerebral blood volume (rCBV(max)) at each habitat with OS. Moreover, the stratification capabilities of the markers to divide patients into "vascular" groups were tested. The variability in the markers between individual centers was also assessed. Statistical Tests Uniparametric Cox regression; Kaplan-Meier test; Mann-Whitney test. Results The rCBV(max) derived from the HAT, LAT, and IPE habitats were significantly associated with patient OS (P < 0.05; hazard ratio [HR]: 1.05, 1.11, 1.28, respectively). Moreover, these markers can stratify patients into "moderate-" and "high-vascular" groups (P < 0.05). The Mann-Whitney test did not find significant differences among most of the centers in markers (HAT: P = 0.02-0.685; LAT: P = 0.010-0.769; IPE: P = 0.093-0.939; VPE: P = 0.016-1.000). Data Conclusion The rCBV(max) calculated in HAT, LAT, and IPE habitats have been validated as clinically relevant prognostic biomarkers for gli

[EN] Objectives To demonstrate how data-driven variability methods can be used to identify changes in disease recording in two English electronic health records databases between 2001 and 2015. Design Repeated cross-sectional analysis that applied data-driven temporal variability methods to assess month-by-month changes in routinely collected medical data. A measure of difference between months was calculated based on joint distributions of age, gender, socioeconomic status and recorded cardiovascular diseases. Distances between months were used to identify temporal trends in data recording. Setting 400 English primary care practices from the Clinical Practice Research Datalink (CPRD GOLD) and 451 hospital providers from the Hospital Episode Statistics (HES). Main outcomes The proportion of patients (CPRD GOLD) and hospital admissions (HES) with a recorded cardiovascular disease (CPRD GOLD: coronary heart disease, heart failure, peripheral arterial disease, stroke; HES: International Classification of Disease codes I20-I69/G45). Results Both databases showed gradual changes in cardiovascular disease recording between 2001 and 2008. The recorded prevalence of included cardiovascular diseases in CPRD GOLD increased by 47%-62%, which partially reversed after 2008. For hospital records in HES, there was a relative decrease in angina pectoris (-34.4%) and unspecified stroke (-42.3%) over the same time period, with a concomitant increase in chronic coronary heart disease (+14.3%). Multiple abrupt changes in the use of myocardial infarction codes in hospital were found in March/April 2010, 2012 and 2014, possibly linked to updates of clinical coding guidelines. Conclusions Identified temporal variability could be related to potentially non-medical causes such as updated coding guidelines. These artificial changes may introduce temporal correlation among diagnoses inferred from routine data, violating the assumptions of frequently used statistical methods. Temporal variabil

[ES] Las asignaturas Sistemas de Información y Telemedicina y Data Quality and Interoperability del Grado y Máster en Ingeniería Biomédica de la Universitat Politècnica de València, España, abordan los resultados de aprendizaje relacionados con la gestión y el procesamiento de bases de datos biomédicas, utilizando estándares de información de salud para la captura e intercambio de datos, evaluación de calidad de datos y desarrollo de modelos de aprendizaje automático a partir de estos datos. Estos resultados de aprendizaje cubren una amplia gama de actividades distintas en el ciclo de vida de los datos biomédicos, lo que puede dificultar el proceso de aprendizaje en el tiempo limitado asignado a la asignatura. Proponemos un enfoque de aprendizaje basado en proyectos que aborde el ciclo de vida completo de los datos biomédicos en el MIMIC-III (Medical Information Mart for Intensive Care III) Open Dataset, una base de datos de libre acceso que comprende información relacionada con pacientes ingresados ​​en unidades de cuidados críticos. Mediante este enfoque de aprendizaje activo, los estudiantes pueden lograr todos los resultados de aprendizaje de la asignatura de manera integrada: comprender el modelo de datos MIMIC-III, usar estándares de información de salud como la Clasificación Internacional de Enfermedades, 9a Edición (CIE-9), mapeo a los estándares de interoperabilidad, consultar datos, crear tablas de datos y abordar la calidad de los datos para aplicar análisis estadísticos y de aprendizaje automático confiables y desarrollar modelos predictivos para varias tareas, como predecir la mortalidad hospitalaria. MIMIC-III es ampliamente utilizado en la academia y la ciencia, con una gran cantidad de recursos disponibles al público y artículos científicos para apoyar el aprendizaje de los estudiantes. Además, los estudiantes obtendrán nuevas competencias en el uso de datos abiertos y capacitación en ética y cumplimiento de la investigación., [EN] The subjects Health Information Systems and Telemedicine and Data Quality and Interoperability of the Degree and Master in Biomedical Engineering of the Universitat Politècnica de València, Spain, address learning outcomes related to managing and processing biomedical databases, using health information standards for data capture and exchange, data quality assessment, and developing machine-learning models from these data. These learning outcomes cover a large range of distinct activities in the biomedical data life-cycle, what may hinder the learning process in the limited time assigned for the subject. We propose a project based learning approach addressing the full life-cycle of biomedical data on the MIMIC-III (Medical Information Mart for Intensive Care III) Open Dataset, a freely accessible database comprising information relating to patients admitted to critical care units. By means of this active learning approach, students can achieve all the learning outcomes of the subject in an integrated manner: understanding the MIMIC-III data model, using health information standards such as International Classification of Diseases 9th Edition (ICD-9), mapping to interoperability standards, querying data, creating data tables and addressing data quality towards applying reliable statistical and machine learning analysis and, developing predictive models for several tasks such as predicting in-hospital mortality. Therefore, three new datasets have been created extracting key information from the MIMIC III database through SQL querys, this data has then been preprocessed to create in-hospital mortality prediction models with the aim of creating an active learning methodology for students and, in a future, implement those models into a clinical decision support system., [CAT/VA] Les assignatures de Sistemes d'Informació i Telemedicina i Data Quality and Interoperability del Grau i Màster d'Enginyeria Biomèdica de la Universitat Politècnica de València, Espanya, aborden els objectius d'aprenentatge relacionats amb el maneig i processament de bases de dades biomèdiques, utilitzant estàndards d'informació en salut per a l'obtenció i l'intercanvi de dades, l'anàlisi de la qualitat d'aquests, i el desenvolupament de models de machine learning. Aquests objectius d'aprenentatge cobreixen un ampli ventall d'activitats diferents en el cicle de vida de les dades biomèdiques, la qual cosa pot dificultar el procés d'aprenentatge en el temps limitat assignat per a cada assignatura. Proposem un aprenentatge basat en projectes abordant el cicle de vida complet de les dades biomèdiques utilitzant el dataset públic MIMIC-III (Medical Information Mart for Intensive Care III), una base de dades accessible gratuïtament comprenent informació relacionada amb l'admissió de pacients en la unitat de vigilància intensiva. Per mitjà d'aquest enfocament d'aprenentatge actiu, els estudiants poden aconseguir tots els objectius d'aprenentatge de les assignatures de manera integral: entenent el model de dades MIMIC-III, utilitzant els estàndards d'informació com International Classification of Diseases 9th Edition (ICD-9), convertint a estàndards d'interoperabilitat, consultant dades, creant taules de dades i abordant la qualitat de dades amb l'objectiu d'aplicar estadístiques fiables i anàlisis machine learning i, desenvolupant múltiples models com els de predicció de mortalitat hospitalària. Per a això, s'han desenvolupat tres nous datasets extraient informació d'importància de la base de dades MIMIC III mitjançant diferents consultes SQL, s'han preprocessat aquestes dades per a un posterior modelatge predictiu de mortalitat hospitalària, amb l'objectiu de crear un aprenentatge actiu per als alumnes i de poder, en un futur, incloure aquests models en un sistema

This is the peer reviewed version of the following article: Pérez-Benito, F.J., Conejero, J.A., Sáez, C., García-Gómez, J.M., Navarro-Pardo, E., Florencio, L.L. and Fernández-de-las-Peñas, C. (2020), Subgrouping Factors Influencing Migraine Intensity in Women: A Semi-automatic Methodology Based on Machine Learning and Information Geometry. Pain Pract, 20: 297-309, which has been published in final form at https://doi.org/10.1111/papr.12854. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Self-Archiving., [EN] Background Migraine is a heterogeneous condition with multiple clinical manifestations. Machine learning algorithms permit the identification of population groups, providing analytical advantages over other modeling techniques. Objective The aim of this study was to analyze critical features that permit the differentiation of subgroups of patients with migraine according to the intensity and frequency of attacks by using machine learning algorithms. Methods Sixty-seven women with migraine participated. Clinical features of migraine, related disability (Migraine Disability Assessment Scale), anxiety/depressive levels (Hospital Anxiety and Depression Scale), anxiety state/trait levels (State-Trait Anxiety Inventory), and pressure pain thresholds (PPTs) over the temporalis, neck, second metacarpal, and tibialis anterior were collected. Physical examination included the flexion-rotation test, cervical range of cervical motion, forward head position while sitting and standing, passive accessory intervertebral movements (PAIVMs) with headache reproduction, and joint positioning sense error. Subgrouping was based on machine learning algorithms by using the nearest neighbors algorithm, multisource variability assessment, and random forest model. Results For migraine intensity, group 2 (women with a regular migraine headache intensity score of 7 on an 11-point Numeric Pain Rating Scale [where 0 = no pain and 10 = maximum pain]) were younger and had lower joint positioning sense error in cervical rotation, greater cervical mobility in rotation and flexion, lower flexion-rotation test scores, positive PAIVMs reproducing migraine, normal PPTs over the tibialis anterior, shorter migraine history, and lower cranio-vertebral angles while standing than the remaining migraine intensity subgroups. The most discriminative variable was the flexion-rotation test score of the symptomatic side. For migraine frequency, no model was able to identify differences between groups (ie, pati

[EN] Background: Temporal variability in health-care processes or protocols is intrinsic to medicine. Such variability can potentially introduce dataset shifts, a data quality issue when reusing electronic health records (EHRs) for secondary purposes. Temporal data-set shifts can present as trends, as well as abrupt or seasonal changes in the statistical distributions of data over time. The latter are particularly complicated to address in multimodal and highly coded data. These changes, if not delineated, can harm population and data-driven research, such as machine learning. Given that biomedical research repositories are increasingly being populated with large sets of historical data from EHRs, there is a need for specific software methods to help delineate temporal data-set shifts to ensure reliable data reuse. Results: EHRtemporalVariability is an open-source R package and Shiny app designed to explore and identify temporal data-set shifts. EHRtemporalVariability estimates the statistical distributions of coded and numerical data over time; projects their temporal evolution through non-parametric information geometric temporal plots; and enables the exploration of changes in variables through data temporal heat maps. We demonstrate the capability of EHRtemporalVariability to delineate data-set shifts in three impact case studies, one of which is available for reproducibility. Conclusions: EHRtemporalVariability enables the exploration and identification of data-set shifts, contributing to the broad examination and repurposing of large, longitudinal data sets. Our goal is to help ensure reliable data reuse for a wide range of biomedical data users. EHRtemporalVariability is designed for technical users who are programmatically utilizing the R package, as well as users who are not familiar with programming via the Shiny user interface.

[EN] Objective To evaluate the effects of Process-Reengineering interventions on the Electronic Health Records (EHR) of a hospital over 7 years. Materials and methods Temporal Variability Assessment (TVA) based on probabilistic data quality assessment was applied to the historic monthly-batched admission data of Hospital La Fe Valencia, Spain from 2010 to 2016. Routine healthcare data with a complete EHR was expanded by processed variables such as the Charlson Comorbidity Index. Results Four Process-Reengineering interventions were detected by quantifiable effects on the EHR: (1) the hospital relocation in 2011 involved progressive reduction of admissions during the next four months, (2) the hospital services re-configuration incremented the number of inter-services transfers, (3) the care-services re-distribution led to transfers between facilities (4) the assignment to the hospital of a new area with 80,000 patients in 2015 inspired the discharge to home for follow up and the update of the pre-surgery planned admissions protocol that produced a significant decrease of the patient length of stay. Discussion TVA provides an indicator of the effect of process re-engineering interventions on healthcare practice. Evaluating the effect of facilities¿ relocation and increment of citizens (findings 1, 3¿4), the impact of strategies (findings 2¿3), and gradual changes in protocols (finding 4) may help on the hospital management by optimizing interventions based on their effect on EHRs or on data reuse. Conclusions The effects on hospitals EHR due to process re-engineering interventions can be evaluated using the TVA methodology. Being aware of conditioned variations in EHR is of the utmost importance for the reliable reuse of routine hospitalization data.

[EN] Depression is associated with absenteeism and presentism, problems in workplace relationships and loss of productivity and quality. The present work describes the validation of a web-based system for the assessment of depression in the university work context. The basis of the system is the Spanish version of the Beck Depression Inventory (BDI-II). A total of 185 participants completed the BDI-II web-based assessment, including 88 males and 97 females, 70 faculty members and 115 staff members. A high level of internal consistency reliability was confirmed. Based on the results of our web-based BDI-II, no significant differences were found in depression severity between gender, age or workers' groups. The main depression risk factors reported were: Changes in sleep, Loss of energy, Tiredness or fatigue and Loss of interest. However significant differences were found by gender in Changes in appetite, Difficulty of concentration and Loss of interest in sex; males expressed less loss of interest in sex than females with a statistically significant difference. Our results indicate that the data collected is coherent with previous BDI-II studies. We conclude that the web-based system based on the BDI-II is psychometrically robust and can be used to assess depression in the university working community.

[EN] Quality of life (QoL) indicators are now being adopted as clinical outcomes in clinical trials on cancer treatments. Technology-free daily monitoring of patients is complicated, time-consuming and expensive due to the need for vast amounts of resources and personnel. The alternative method of using the patients¿ own phones could reduce the burden of continuous monitoring of cancer patients in clinical trials. This paper proposes monitoring the patients¿ QoL by gathering data from their own phones. We considered that the continuous multiparametric acquisition of movement, location, phone calls, conversations and data use could be employed to simultaneously monitor their physical, psychological, social and environmental aspects. An open access phone app was developed (Human Dynamics Reporting Service (HDRS)) to implement this approach. We here propose a novel mapping between the standardized QoL items for these patients, the European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) and define HDRS monitoring indicators. A pilot study with university volunteers verified the plausibility of detecting human activity indicators directly related to QoL.

Los datos se han convertido en un torrente que fluye en todos los ámbitos de la sociedad. La aplicación de la Inteligencia Artificial al ámbito de la salud es posiblemente el de mayor impacto social, por las implicaciones que tiene para el bienestar de las personas. Muchos han depositado las esperanzas de la Medicina de Precisión en los Sistemas de Ayuda a la Decisión Médica, objeto de estudio de este libro El desarrollo de Sistemas de Ayuda a la Decisión Médica abarca múltiples aspectos procedentes de la medicina, la Inteligencia Artificial, la teoría de la decisión, la ingeniería biomédica, la biología, la farmacia, las organizaciones sanitarias,la ingeniería de procesos, la evaluación de sistemas y el sentido común. En este libro, elaboraremos los conceptos estratégicos, funcionales y técnicos necesarios para concebir, diseñar, adaptar, implantar y evaluar Sistemas de Ayuda a la Decisión Médica. Los textos están dirigidos a alumnos y profesionales de Tecnologías de Información para la Salud, Medicina, e Ingeniería Biomédica aprovechando casos reales de proyectos desarrollados por los propios autores

El cáncer de mama es el tumor maligno más frecuente entre las mujeres de todo el mundo, (a excepción de los tumores de cánceres de piel). Se estima que el riesgo de padecer cáncer de mama es de, aproximadamente, 1 de cada 8 mujeres. El cáncer de mama no invasivo más común es el carcinoma ductal in situ (CDIS). El CDIs es un precursor del cáncer de mama invasivo, pero no siempre evoluciona en esa dirección Aunque existen guías clínicas definidas para el tratamiento de esta enfermedad, aparecen ciertas situaciones en las que los pasos a seguir no están completamente definidos, constituyendo sucesos inciertos en las diferentes etapas de sus rutas asistenciales. El objetivo de este trabajo es diseñar un árbol de decisión Bayesiana para el carcinoma ductal in situ de mama, cuya representación cronológica del proceso de decisión permita proporcionar un soporte cuantitativo a los expertos que determinen el tratamiento. Para el desarrollo de esta herramienta se ha tomado como base la Teoría de la Toma de Decisiones y se han obtenido las opciones de tratamiento óptimas, a partir del árbol de decisión diseñado, para cuatro poblaciones de pacientes, diferenciadas por unas características determinadas, que resultan de interés para determinar la evolución del CDIS., Breast cancer is the most common malignant tumor among women worldwide (except for skin cancer tumors). The estimated risk of breast cancer is about 1 in 8 women. The most common non-invasive breast cancer is ductal carcinoma in situ (DCIS). DCIS are a precursor to invasive breast cancer, but they do not always progress in that direction. Although there are clinical guidelines for the treatment of this disease, certain situations arise in which the steps to be followed are not completely defined, constituting uncertain events in the different stages of their care routes. The aim of this paper is to design a Bayesian decision tree for ductal carcinoma in situ, whose chronological representation of the decision process will provide quantitative support to the experts who determine the treatment. For the development of this tool, the Theory of Decision Making has been taken as the basis and the optimal treatment options have been obtained from the decision tree designed for four patient populations, differentiated by certain characteristics, which are of interest for determining the evolution of the CDIS., El càncer de mama és el tumor maligne més freqüent entre les dones de tot el món. S'estima que el risc de patir càncer de mama és de, aproximadament, 1 de cada 8 dones. El càncer de mama no invasiu més comú és el carcinoma ductal in situ (CDIS). El CDIS és un precursor del càncer de mama invasiu, però no sempre evoluciona en eixa direcció. Encara que existixen guies clíniques definides per al tractament d'esta malaltia, apareixen certes situacions en què els passos que s'ha de seguir no estan completament definits, constituint successos incerts en les diferents etapes de les seues rutes assistencials. L'objectiu d'este Treball Final de Grau és dissenyar un arbre de decisió Bayesiana per al carcinoma ductal in situ de mamella, la representació cronològica del qual procés de decisió permeta proporcionar un suport quantitatiu als experts que determinen el tractament. Per al desenrotllament d'esta ferramenta s'ha pres com a base la Teoria de La Presa de Decisions i s'han obtingut les opcions de tractament òptimes, a partir de l'arbre de decisió dissenyat, per a quatre poblacions de pacients, diferenciades per unes característiques determinades, que resulten d'interés per a determinar l'evolució del CDIS.

[EN] The aim of this work was to develop a new unsupervised exploratory method of characterizing feature extraction and detecting similarity of movement during sleep through actigraphy signals. We here propose some algorithms, based on signal bispectrum and bispectral entropy, to determine the unique features of independent actigraphy signals. Experiments were carried out on 20 randomly chosen actigraphy samples of the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) database, with no information other than their aperiodicity. The Pearson correlation coefficient matrix and the histogram correlation matrix were computed to study the similarity of movements during sleep. The results obtained allowed us to explore the connections between certain sleep actigraphy patterns and certain pathologies.

The degree of homogeneity among data distributions is a critical issue when reusing data integrated from different sources, since the introduction of undesired variabilities may lead to misleading results. Therefore, assessing this data source variability is of utmost importance to ensure a confident data reuse. In the biomedical field, this issue is even more critical, due to the risk of serious consequences if data is used improperly. Furthermore, in spite of the existence of classical statistical tools which attempt to deal with this task, there are still several aspects to be addressed due to inherent complexity of biomedical data, such as the intrinsic multimodality of data distributions. New statistical metrics have been recently developed to overcome this challenge, including the Global probabilistic deviation (GPD) and the Source probabilistic outlyingness (SPO). These metrics allow a multivariate analysis of data source variability without assuming any underlying distribution and without being restricted to certain types of data. However, when implementing them, users must decide among different options related to data preprocessing techniques, as data representation, reduction and normalization. This fact is not a limitation, but needs to be addressed. In this work, an in-depth study of the influence of preprocessing techniques over the multi-source variability metrics is performed, and procedures to overcome the problematic phenomena encountered are proposed and evaluated. Once understood the influence of the different settings, the potential biases introduced by pre-set factors to the GPD and SPO metrics, such as the number of sources or the number of data, are evaluated. Results of this assessment suggest the robustness of GPD and SPO to these factors. Finally, new procedures are proposed to find patterns in multi-source biomedical repositories and offer suggestions for data reuse based on the different data source variability structures. A new clusteri, El grado de homogeneidad entre distribuciones de datos es una circunstancia crítica cuando se reutilizan datos provenientes de diferentes fuentes, ya que la introducción de variabilidades no deseadas podría conducir a resultados engañosos. Por lo tanto, evaluar esta variabilidad producida por la fuente de donde provienen los datos es clave de cara a asegurar una reutilización segura de los mismos. En el ámbito biomédico, este problema es aún más importante, debido al riesgo de consecuencias graves si los datos son utilizados de forma inadecuada. Además, a pesar de la existencia de herramientas estadísticas clásicas que intentar afrontar esta tarea, todavía existen varios aspectos que requieren ser tratados, fruto de la inherente complejidad de los datos biomédicos. Nuevas métricas estadísticas han sido recientemente desarrolladas para afrontar este reto, incluyendo la Global probabilistic deviation (GPD) y la Source probabilistic outlyingness (SPO). Estas métricas permiten un análisis multivariante de la variabilidad introducida por la fuente de datos sin asumir ninguna distribución subyacente y sin estar restringidas a ciertos tipos de datos. Sin embargo, a la hora de implementarlas, los usuarios tienen que decidir entre diferentes opciones relacionadas con técnicas de preprocesamiento. Esta circunstancia no supone una limitación, pero debe ser abordada. En este trabajo, se lleva a cabo un estudio en profundidad de la influencia de las técnicas de preprocesamiento sobre las métricas de variabilidad multi fuente, y se proponen procedimientos para superar los fenómenos adversos encontrados. Una vez entendida la influencia de las diferentes configuraciones, se evalúan los potenciales sesgos introducidos por factores preestablecidos, sobre la GPD y la SPO, tales como el número de fuentes o el número de datos. Los resultados de este análisis sugieren la robustez de GPD y SPO frente estos factores. Finalmente, se proponen nuevos procedimientos para encontrar patrones en, El grau d’homogeneïtat entre distribucions de dades és una circumstància crítica quan es reutilitzen dades provinents de diferents fonts, ja que la introducció de variabilitats no desitjades podria conduir a resultats enganyosos. Per tant, avaluar aquesta variabilitat produïda per la font d’on provenen les dades esdevé clau de cara a assegurar una reutilització segura d’aquestes. En l’àmbit biomèdic, aquest problema és encara més important, a causa del risc de conseqüències greus si les dades son utilitzades de forma inadequada. A més, a pesar de l’existència d’eines estadístiques clàssiques que intentar afrontar aquesta tasca, encara existeixen diversos aspectes que requereixen ser tractats i que són fruit de la inherent complexitat de les dades biomèdiques. Noves mètriques estadístiques han sigut recentment desenvolupades per a afrontar aquest repte, incloent la Global probabilistic deviation (GPD) i la Source probabilistic outlyingness (SPO). Aquestes mètriques permeten una anàlisi multivariant de la variabilitat de la font de dades sense assumir cap distribució subjacent i sense estar restringides a certs tipus de dades. Tanmateix, a l’hora d’implementar-les, els usuaris han de decidir entre diferents opcions relacionades amb tècniques de preprocessament. Aquesta circumstància no suposa una limitació, però requereix ser tractada. En aquest treball, es du a terme un estudi en profunditat de la influència de les tècniques de preprocessament sobre les mètriques de variabilitat multi font, i es proposen procediments per a superar els fenòmens adversos trobats. Una vegada entesa la influència de les diferents configuracions, s'avaluen els potencials biaixos introduïts per factors preestablerts, sobre la GPD i la SPO, com ara el nombre de fonts o el nombre de dades. Els resultats d'aquesta anàlisis suggereixen la robustesa de GPD i SPO front aquests factors. Finalment, es proposen nous procediments per a trobar patrons en repositoris biomèdics multi font, així com sug

[EN] This is the second in a series of two papers regarding the construction of data quality (DQ) assured repositories, based on population data from Electronic Health Records (EHR), for the reuse of information on infant feeding from birth until the age of two. This second paper describes the application of the computational process of constructing the first quality-assured repository for the reuse of information on infant feeding in the perinatal period, with the aim of studying relevant questions from the Baby Friendly Hospital Initiative (BFHI) and monitoring its deployment in our hospital. The construction of the repository was carried out using 13 semi-automated procedures to assess, recover or discard clinical data. The initial information consisted of perinatal forms from EHR related to 2048 births (Facts of Study, FoS) between 2009 and 2011, with a total of 433,308 observations of 223 variables. DQ was measured before and after the procedures using metrics related to eight quality dimensions: predictive value, correctness, duplication, consistency, completeness, contextualization, temporal-stability, and spatial-stability. Once the predictive variables were selected and DQ was assured, the final repository consisted of 1925 births, 107,529 observations and 73 quality-assured variables. The amount of discarded observations mainly corresponds to observations of non-predictive variables (52.90%) and the impact of the de-duplication process (20.58%) with respect to the total input data. Seven out of thirteen procedures achieved 100% of valid births, observations and variables. Moreover, 89% of births and 98% of observations were consistent according to the experts' criteria. A multidisciplinary approach along with the quantification of DQ has allowed us to construct the first repository about infant feeding in the perinatal period based on EHR population data.

[EN] Nowadays, biomedical research and decision making depend to a great extent on the data stored in information systems. As a consequence, a lack of data quality (DQ) may lead to suboptimal decisions, or hinder the derived research processes and outcomes. This thesis aims to the research and development of methods for assessing two DQ problems of special importance in Big Data and large-scale repositories, based on multi-institutional, cross-border infrastructures, and acquired during long periods of time: the variability of data probability distributions (PDFs) among different data sources-multi-source variability-and the variability of data PDFs over time-temporal variability. Variability in PDFs may be caused by differences in data acquisition methods, protocols or health care policies; systematic or random errors during data input and management; demographic differences in populations; or even falsified data. To date, these issues have received little attention as DQ problems nor count with adequate assessment methods. The developed methods aim to measure, detect and characterize variability dealing with multi-type, multivariate, multi-modal data, and not affected by large sample sizes. To this end, we defined an Information Theory and Geometry probabilistic framework based on the inference of non-parametric statistical manifolds from the normalized distances of PDFs among data sources and over time. Based on this, a number of contributions have been generated. For the multi-source variability assessment we have designed two metrics: the Global Probabilistic Deviation, which measures the degree of global variability among the PDFs of multiple sources-equivalent to the standard deviation among PDFs; and the Source Probabilistic Outlyingness, which measures the dissimilarity of the PDF of a single data source to a global latent average. They are based on the construction of a simplex geometrical figure (the maximum-dimensional statistical manifold) using the dis, [ES] Actualmente, la investigación biomédica y toma de decisiones dependen en gran medida de los datos almacenados en los sistemas de información. En consecuencia, una falta de calidad de datos (CD) puede dar lugar a decisiones sub-óptimas o dificultar los procesos y resultados de las investigaciones derivadas. Esta tesis tiene como propósito la investigación y desarrollo de métodos para evaluar dos problemas especialmente importantes en repositorios de datos masivos (Big Data), basados en infraestructuras multi-céntricas, adquiridos durante largos periodos de tiempo: la variabilidad de las distribuciones de probabilidad (DPs) de los datos entre diferentes fuentes o sitios-variabilidad multi-fuente-y la variabilidad de las distribuciones de probabilidad de los datos a lo largo del tiempo-variabilidad temporal. La variabilidad en DPs puede estar causada por diferencias en los métodos de adquisición, protocolos o políticas de atención; errores sistemáticos o aleatorios en la entrada o gestión de datos; diferencias demográficas en poblaciones; o incluso por datos falsificados. Esta tesis aporta métodos para detectar, medir y caracterizar dicha variabilidad, tratando con datos multi-tipo, multivariantes y multi-modales, y sin ser afectados por tamaños muestrales grandes. Para ello, hemos definido un marco de Teoría y Geometría de la Información basado en la inferencia de variedades de Riemann no-paramétricas a partir de distancias normalizadas entre las PDs de varias fuentes de datos o a lo largo del tiempo. En consecuencia, se han aportado las siguientes contribuciones: Para evaluar la variabilidad multi-fuente se han definido dos métricas: la Global Probabilistic Deviation, la cual mide la variabilidad global entre las PDs de varias fuentes-equivalente a la desviación estándar entre PDs; y la Source Probabilistic Outlyingness, la cual mide la disimilaridad entre la DP de una fuente y un promedio global latente. Éstas se basan en un simplex construido mediante las dist, [CA] Actualment, la investigació biomèdica i presa de decisions depenen en gran mesura de les dades emmagatzemades en els sistemes d'informació. En conseqüència, una manca en la qualitat de les dades (QD) pot donar lloc a decisions sub-òptimes o dificultar els processos i resultats de les investigacions derivades. Aquesta tesi té com a propòsit la investigació i desenvolupament de mètodes per avaluar dos problemes especialment importants en repositoris de dades massius (Big Data) basats en infraestructures multi-institucionals o transfrontereres, adquirits durant llargs períodes de temps: la variabilitat de les distribucions de probabilitat (DPs) de les dades entre diferents fonts o llocs-variabilitat multi-font-i la variabilitat de les distribucions de probabilitat de les dades al llarg del temps-variabilitat temporal. La variabilitat en DPs pot estar causada per diferències en els mètodes d'adquisició, protocols o polítiques d'atenció; errors sistemàtics o aleatoris durant l'entrada o gestió de dades; diferències demogràfiques en les poblacions; o fins i tot per dades falsificades. Aquesta tesi aporta mètodes per detectar, mesurar i caracteritzar aquesta variabilitat, tractant amb dades multi-tipus, multivariants i multi-modals, i no sent afectats per mides mostrals grans. Per a això, hem definit un marc de Teoria i Geometria de la Informació basat en la inferència de varietats de Riemann no-paramètriques a partir de distàncies normalitzades entre les DPs de diverses fonts de dades o al llarg del temps. En conseqüència s'han aportat les següents contribucions: Per avaluar la variabilitat multi-font s'han definit dos mètriques: la Global Probabilistic Deviation, la qual mesura la variabilitat global entre les DPs de les diferents fonts-equivalent a la desviació estàndard entre DPs; i la Source Probabilistic Outlyingness, la qual mesura la dissimilaritat entre la DP d'una font de dades donada i una mitjana global latent. Aquestes estan basades en la construcció d'un

[CA] Aquest projecte consta de la implementació d’una prova de concepte per a demostrar l’us de les tecnologies Big Data a l’actualitat a l’àmbit de la salut. S’han preparat les diverses part del programa pensant en que siga accessible via web, altament escalable, fiable, i útil per al sistema sanitari actual. En concret es centra amb el framework de Storm, basat en Streaming i al que li se dona utilitat en molts àmbits en l’actualitat; la seua filosofia de Streams encaixa de forma molt adequada amb el problema de les rutes assistencials proposat al títol, i dona moltes possibilitats al tractament d’aquestes de forma rapida i distribuïda.

OBJECTIVE: To assess the variability in data distributions among data sources and over time through a case study of a large multisite repository as a systematic approach to data quality (DQ). MATERIALS AND METHODS: Novel probabilistic DQ control methods based on information theory and geometry are applied to the Public Health Mortality Registry of the Region of Valencia, Spain, with 512 143 entries from 2000 to 2012, disaggregated into 24 health departments. The methods provide DQ metrics and exploratory visualizations for (1) assessing the variability among multiple sources and (2) monitoring and exploring changes with time. The methods are suited to big data and multitype, multivariate, and multimodal data. RESULTS: The repository was partitioned into 2 probabilistically separated temporal subgroups following a change in the Spanish National Death Certificate in 2009. Punctual temporal anomalies were noticed due to a punctual increment in the missing data, along with outlying and clustered health departments due to differences in populations or in practices. DISCUSSION: Changes in protocols, differences in populations, biased practices, or other systematic DQ problems affected data variability. Even if semantic and integration aspects are addressed in data sharing infrastructures, probabilistic variability may still be present. Solutions include fixing or excluding data and analyzing different sites or time periods separately. A systematic approach to assessing temporal and multisite variability is proposed. CONCLUSION: Multisite and temporal variability in data distributions affects DQ, hindering data reuse, and an assessment of such variability should be a part of systematic DQ procedures.

This is the first paper of a series of two regarding the construction of data quality (DQ) assured repositories for the reuse of information on infant feeding from birth until two years old. This first paper justifies the need for such repositories and describes the design of a process to construct them from Electronic Health Records (EHR). As a result, Part 1 proposes a computational process to obtain qualityassured datasets represented by a canonical structure extracted from raw data from multiple EHR. For this, 13 steps were defined to ensure the harmonization, standardization, completion, de-duplication, and consistency of the dataset content. Moreover, the quality of the input and output data for each of these steps is controlled according to eight DQ dimensions: predictive value, correctness, duplication, consistency, completeness, contextualization, temporal-stability and spatial-stability. The second paper of the series will describe the application of this computational process to construct the first qualityassured repository for the reuse of information on infant feeding in the perinatal period aimed at the monitoring of clinical activities and research.

[EN] Biomedical data may be composed of individuals generated from distinct, meaningful sources. Due to possible contextual biases in the processes that generate data, there may exist an undesirable and unexpected variability among the probability distribution functions (PDFs) of the source subsamples, which, when uncontrolled, may lead to inaccurate or unreproducible research results. Classical statistical methods may have difficulties to undercover such variabilities when dealing with multi-modal, multi-type, multi-variate data. This work proposes two metrics for the analysis of stability among multiple data sources, robust to the aforementioned conditions, and defined in the context of data quality assessment. Specifically, a global probabilistic deviation (GPD) and a source probabilistic outlyingness (SPO) metrics are proposed. The first provides a bounded degree of the global multi-source variability, designed as an estimator equivalent to the notion of normalized standard deviation of PDFs. The second provides a bounded degree of the dissimilarity of each source to a latent central distribution. The metrics are based on the projection of a simplex geometrical structure constructed from the Jensen-Shannon distances among the sources PDFs. The metrics have been evaluated and demonstrated their correct behaviour on a simulated benchmark and with real multi-source biomedical data using the UCI Heart Disease dataset. The biomedical data quality assessment based on the proposed stability metrics may improve the efficiency and effectiveness of biomedical data exploitation and research.

The final publication is available at Springer via http://dx.doi.org/DOI 10.1007/s10618-014-0378-6. Published online., Knowledge discovery on biomedical data can be based on on-line, data-stream analyses, or using retrospective, timestamped, off-line datasets. In both cases, changes in the processes that generate data or in their quality features through time may hinder either the knowledge discovery process or the generalization of past knowledge. These problems can be seen as a lack of data temporal stability. This work establishes the temporal stability as a data quality dimension and proposes new methods for its assessment based on a probabilistic framework. Concretely, methods are proposed for (1) monitoring changes, and (2) characterizing changes, trends and detecting temporal subgroups. First, a probabilistic change detection algorithm is proposed based on the Statistical Process Control of the posterior Beta distribution of the Jensen–Shannon distance, with a memoryless forgetting mechanism. This algorithm (PDF-SPC) classifies the degree of current change in three states: In-Control, Warning, and Out-of-Control. Second, a novel method is proposed to visualize and characterize the temporal changes of data based on the projection of a non-parametric information-geometric statistical manifold of time windows. This projection facilitates the exploration of temporal trends using the proposed IGT-plot and, by means of unsupervised learning methods, discovering conceptually-related temporal subgroups. Methods are evaluated using real and simulated data based on the National Hospital Discharge Survey (NHDS) dataset.

La falta de garantías en la calidad de los datos en el ámbito biomédico puede llevar a tomar malas decisiones en el tratamiento de los pacientes y conclusiones erróneas en los estudios poblacionales. Para contribuir a resolver este problema, presentamos el desarrollo de un servicio web implementado en Java para medir la calidad de los datos biomédicos, conjuntamente con una interfaz accesible desde cualquier navegador web para poder hacer uso del servicio. Para la medición de la calidad, se ha empleado una aproximación resultante de combinar nueve dimensiones de calidad (Completeness, Consistency, Duplication, Correctness, Temporal Stability, Spatial Stability, y Contextualization), aplicadas a distintos ejes de los datos (Dataset, Attribute, Registry, Value, Source y Time). La arquitectura está diseñada según el patrón estrategia, que permite la implementación del método de medida de cada dimensión sobre la clase abstracta 'Axis', que se especializa en los seis posibles ejes. Como prueba de concepto, se ha procedido a la implementación de una dimensión de conteo llamada completitud, y otra basada en reglas, llamada consistencia. Las tecnologías utilizadas para poder realizar el proyecto han sido principalmente el entorno de desarrollo integrado Eclipse, el sistema de gestión de reglas de negocio Drools, el conjunto de herramientas para la implementación de servicios web Jersey, y el framework para la creación de interfaces gráficas de usuario web Vaadin. En el desarrollo se han respetado las condiciones de escalabilidad y rendimiento, que requiere un proyecto destinado al manejo de un volumen de datos que puede llegar a ser elevado. Además, se ha logrado realizar todo el proyecto empleando bibliotecas de código libre.

Research biobanks are often composed by data from multiple sources. In some cases, these different subsets of data may present dissimilarities among their probability density functions (PDF) due to spatial shifts. This, may lead to wrong hypothesis when treating the data as a whole. Also, the overall quality of the data is diminished. With the purpose of developing a generic and comparable metric to assess the stability of multi-source datasets, we have studied the applicability and behaviour of several PDF distances over shifts on different conditions (such as uni- and multivariate, different types of variable, and multi-modality) which may appear in real biomedical data. From the studied distances, we found information-theoretic based and Earth Mover’s Distance to be the most practical distances for most conditions. We discuss the properties and usefulness of each distance according to the possible requirements of a general stability metric.

The success of Clinical Decision Support Systems (CDSS) greatly depends on its capability of being integrated in Health Information Systems (HIS). Several proposals have been published up to date to permit CDSS gathering patient data from HIS. Some base the CDSS data input on the HL7 reference model, however, they are tailored to specific CDSS or clinical guidelines technologies, or do not focus on standardizing the CDSS resultant knowledge. We propose a solution for facilitating semantic interoperability to rule-based CDSS focusing on standardized input and output documents conforming an HL7-CDA wrapper. We define the HL7-CDA restrictions in a HL7-CDA implementation guide. Patient data and rule inference results are mapped respectively to and from the CDSS by means of a binding method based on an XML binding file. As an independent clinical document, the results of a CDSS can present clinical and legal validity. The proposed solution is being applied in a CDSS for providing patient-specific recommendations for the care management of outpatients with diabetes mellitus.

[EN] New biomedical technologies enable the diagnosis of brain tumours by using non-invasive methods. HealthAgents is a European Union-funded research project that aims to build an agent-based distributed decision support system (dDSS) for the diagnosis of brain tumours. This is achieved using the latest biomedical knowledge, information and communication technologies and pattern recognition (PR) techniques. As part of the PR development of HealthAgents, an independent and automatic classification framework (CF) has been developed. This framework has been integrated with the HealthAgents dDSS using the HealthAgents agent platform. The system offers (1) the functionality to search for distributed classifiers to solve specific questions; (2) automatic classification of new cases; (3) instant deployment of new validated classifiers; and (4) the ability to rank a set of classifiers according to their performance and suitability for the case in hand. The CF enables both the deployment of new classifiers using the provided Extensible Markup Language(1) classifier specification, and the inclusion of new PR techniques that make the system extensible. These features may enable the rapid integration of PR laboratory results into industrial or research applications, such as the HealthAgents dDSS. Two classification nodes have been deployed and they currently offer classification services by means of dedicated servers connected to the HealthAgents agent platform: one node being located at the Katholieke Universiteit Leuven, Belgium and the other at the Universidad Politecnica de Valencia, Spain. These classification nodes share the current set of brain tumour classifiers that have been trained from in vivo magnetic resonance spectroscopy data. The combination of the CF with a distributed agent system constitutes the basis of the brain tumour dDSS developed in HealthAgents.

Los Sistemas de Ayuda a la Decisión Clínica (CDSS) actuales son juzgados por su flexibilidad y adaptabilidad. Se describe el desarrollo de un CDSS distribuido basado en agentes, un motor de clasificación basado en reconocimiento de formas, y el DSS genérico Curiam. En conjunto ofrecen una solución reutilizable para el desarrollo de nuevos CDSS.

In this paper we present our experience of representing the knowledge behind HealthAgents (HA), a distributed decision support system for brain tumour diagnosis. Our initial motivation came from the distributed nature of the information involved in the system and has been enriched by clinicians' requirements and data access restrictions. We present in detail the steps we have taken towards building our ontology starting from knowledge acquisition to data access and reasoning. We motivate our representational choices and show our results using domain examples used by clinical partners in HA. © Copyright Cambridge University Press 2011.

The final publication is available at Springer via http://dx.doi.org/10.1007/s10334-010-0241-8, Object: This study demonstrates that 3T SV-MRS data can be used with the currently available automatic brain tumour diagnostic classifiers which were trained on databases of 1.5T spectra. This will allow the existing large databases of 1.5T MRS data to be used for diagnostic classification of 3T spectra, and perhaps also the combination of 1.5T and 3T databases. Materials and methods: Brain tumour classifiers trained with 154 1.5T spectra to discriminate among high grade malignant tumours and common grade II glial tumours were evaluated with a subsequently-acquired set of 155 1.5T and 37 3T spectra. A similarity study between spectra and main brain tumour metabolite ratios for both field strengths (1.5T and 3T) was also performed. Results: Our results showed that classifiers trained with 1.5T samples had similar accuracy for both test datasets (0.87 ± 0.03 for 1.5T and 0.88 ± 0.03 for 3.0T). Moreover, non-significant differences were observed with most metabolite ratios and spectral patterns. Conclusion: These results encourage the use of existing classifiers based on 1.5T datasets for diagnosis with 3T 1H SV-MRS. The large 1.5T databases compiled throughout many years and the prediction models based on 1.5T acquisitions can therefore continue to be used with data from the new 3T instruments. © 2011 ESMRMB.