Collective intelligence

Estudio e implantación de algoritmos de recomendación, búsqueda, ranking y aprendizaje.

mpiGL : un projecte d'utilització del programari lliure en la visualització gràfica en xarxes de càlcul científic

S'analitzen les problemàtiques relacionades amb la presentació d'informació gràfica en temps real durant un càlcul paral·lel o col·laboratiu en un entorn distribuït, i es fa una proposta de toolkit obert que estén el llenguatge OpenGL per la seva resolució.

Prediction of postpartum depression using multilayer perceptrons and pruning.

OBJECTIVE. The main goal of this paper is to obtain a classification model based on feed-forward multilayer perceptrons in order to improve postpartum depression prediction during the 32 weeks after childbirth with a high sensitivity and specificity and to develop a tool to be integrated in a decision support system for clinicians. MATERIALS AND METHODS. Multilayer perceptrons were trained on data from 1397 women who had just given birth, from seven Spanish general hospitals, including clinical, environmental and genetic variables. A prospective cohort study was made just after delivery, at 8 weeks and at 32 weeks after delivery. The models were evaluated with the geometric mean of accuracies using a hold-out strategy. RESULTS. Multilayer perceptrons showed good performance (high sensitivity and specificity) as predictive models for postpartum depression. CONCLUSIONS. The use of these models in a decision support system can be clinically evaluated in future work. The analysis of the models by pruning leads to a qualitative interpretation of the influence of each variable in the interest of clinical protocols.

Assessment of drug-induced hepatotoxicity in clinical practice: a challenge for gastroenterologists.

Currently, pharmaceutical preparations are serious contributors to liver disease; hepatotoxicity ranking as the most frequent cause for acute liver failure and post-commercialization regulatory decisions. The diagnosis of hepatotoxicity remains a difficult task because of the lack of reliable markers for use in general clinical practice. To incriminate any given drug in an episode of liver dysfunction is a step-by-step process that requires a high degree of suspicion, compatible chronology, awareness of the drug's hepatotoxic potential, the exclusion of alternative causes of liver damage and the ability to detect the presence of subtle data that favors a toxic etiology. This process is time-consuming and the final result is frequently inaccurate. Diagnostic algorithms may add consistency to the diagnostic process by translating the suspicion into a quantitative score. Such scales are useful since they provide a framework that emphasizes the features that merit attention in cases of suspected hepatic adverse reaction as well. Current efforts in collecting bona fide cases of drug-induced hepatotoxicity will make refinements of existing scales feasible. It is now relatively easy to accommodate relevant data within the scoring system and to delete low-impact items. Efforts should also be directed toward the development of an abridged instrument for use in evaluating suspected drug-induced hepatotoxicity at the very beginning of the diagnosis and treatment process when clinical decisions need to be made. The instrument chosen would enable a confident diagnosis to be made on admission of the patient and treatment to be fine-tuned as further information is collected.

Aplicació de visió per dispositius mòbils

En aquest projecte es presenta l’aplicació per a dispositius mòbils Doppelganger. La seva funció és, a partir d’una fotografia, detectar la cara i mostrar la persona famosa de la nostra base de dades que més s’assembla a la persona en la fotografia. Per la implementació s’han utilitzat algoritmes de visió per computador i d’aprenentatge automàtic com per exemple el PCA i el K-Nearest Neighbor, tot utilitzant llibreries gratuïtes com són les OpenCV.

Optimització i personalització del càlcul de rutes en dispositius mòbils

Aquest estudi es centra en trobar la ruta que millor satisfaci les necessitats i preferències de l'usuari (personalització). Així, partint d'algorismes de routing existents, caldrà estudiar quin és el millor algorisme per implementar la funció de cost que doni la "bellesa" de la ruta per l'usuari.

Estudi comparatiu de l'algorisme basat en una colònia artificial d'abelles

La finalitat d'aquest projecte és la realització d'un estudi comparatiu de l'algoritme basat en una colònia artificial d'abelles, Artificial Bee Colony (ABC), comparat amb un conjunt d'algoritmes fonamentats en el paradigma de la computació evolutiva. S'utilitzarà l'eficàcia a l'hora d'optimitzar diverses funcions com a mesura comparativa. Els algoritmes amb els quals es comparara l'algoritme ABC són: algoritmes genètics, evolució diferencial i optimització amb eixam de partícules.

Learning about Distributed Systems

This paper presents the "state of the art" about distributed systems and applications and it's focused on teaching about these systems. It presents different platforms where to run distributed applications and describes some development toolkits whose can be used to develop prototypes, practices and distributed applications. It also presents some existing distributed algorithms useful for class practices, and some tools to help managing distributed environments. Finally, the paper presents some teaching experiences with different approaches on how to teach about distributed systems.

Route optimization and customization using real geographical data in Android mobile devices

Nowadays, there are several services and applications that allow users to locate and move to different tourist areas using a mobile device. These systems can be used either by internet or downloading an application in concrete places like a visitors centre. Although such applications are able to facilitate the location and the search for points of interest, in most cases, these services and applications do not meet the needs of each user. This paper aims to provide a solution by studying the main projects, services and applications, their routing algorithms and their treatment of the real geographical data in Android mobile devices, focusing on the data acquisition and treatment to improve the routing searches in off-line environments.

Using free data mining software and clustering algorithms to find predictors from student qualifications

In this project a research both in finding predictors via clustering techniques and in reviewing the Data Mining free software is achieved. The research is based in a case of study, from where additionally to the KDD free software used by the scientific community; a new free tool for pre-processing the data is presented. The predictors are intended for the e-learning domain as the data from where these predictors have to be inferred are student qualifications from different e-learning environments. Through our case of study not only clustering algorithms are tested but also additional goals are proposed.

Effective diagnosis of Alzheimer's disease by means of large margin-based methodology.

BACKGROUND Functional brain images such as Single-Photon Emission Computed Tomography (SPECT) and Positron Emission Tomography (PET) have been widely used to guide the clinicians in the Alzheimer's Disease (AD) diagnosis. However, the subjectivity involved in their evaluation has favoured the development of Computer Aided Diagnosis (CAD) Systems. METHODS It is proposed a novel combination of feature extraction techniques to improve the diagnosis of AD. Firstly, Regions of Interest (ROIs) are selected by means of a t-test carried out on 3D Normalised Mean Square Error (NMSE) features restricted to be located within a predefined brain activation mask. In order to address the small sample-size problem, the dimension of the feature space was further reduced by: Large Margin Nearest Neighbours using a rectangular matrix (LMNN-RECT), Principal Component Analysis (PCA) or Partial Least Squares (PLS) (the two latter also analysed with a LMNN transformation). Regarding the classifiers, kernel Support Vector Machines (SVMs) and LMNN using Euclidean, Mahalanobis and Energy-based metrics were compared. RESULTS Several experiments were conducted in order to evaluate the proposed LMNN-based feature extraction algorithms and its benefits as: i) linear transformation of the PLS or PCA reduced data, ii) feature reduction technique, and iii) classifier (with Euclidean, Mahalanobis or Energy-based methodology). The system was evaluated by means of k-fold cross-validation yielding accuracy, sensitivity and specificity values of 92.78%, 91.07% and 95.12% (for SPECT) and 90.67%, 88% and 93.33% (for PET), respectively, when a NMSE-PLS-LMNN feature extraction method was used in combination with a SVM classifier, thus outperforming recently reported baseline methods. CONCLUSIONS All the proposed methods turned out to be a valid solution for the presented problem. One of the advances is the robustness of the LMNN algorithm that not only provides higher separation rate between the classes but it also makes (in combination with NMSE and PLS) this rate variation more stable. In addition, their generalization ability is another advance since several experiments were performed on two image modalities (SPECT and PET).

Selection of Reference Genes for Quantitative Real Time PCR (qPCR) Assays in Tissue from Human Ascending Aorta.

Dilatation of the ascending aorta (AAD) is a prevalent aortopathy that occurs frequently associated with bicuspid aortic valve (BAV), the most common human congenital cardiac malformation. The molecular mechanisms leading to AAD associated with BAV are still poorly understood. The search for differentially expressed genes in diseased tissue by quantitative real-time PCR (qPCR) is an invaluable tool to fill this gap. However, studies dedicated to identify reference genes necessary for normalization of mRNA expression in aortic tissue are scarce. In this report, we evaluate the qPCR expression of six candidate reference genes in tissue from the ascending aorta of 52 patients with a variety of clinical and demographic characteristics, normal and dilated aortas, and different morphologies of the aortic valve (normal aorta and normal valve n = 30; dilated aorta and normal valve n = 10; normal aorta and BAV n = 4; dilated aorta and BAV n = 8). The expression stability of the candidate reference genes was determined with three statistical algorithms, GeNorm, NormFinder and Bestkeeper. The expression analyses showed that the most stable genes for the three algorithms employed were CDKN1β, POLR2A and CASC3, independently of the structure of the aorta and the valve morphology. In conclusion, we propose the use of these three genes as reference genes for mRNA expression analysis in human ascending aorta. However, we suggest searching for specific reference genes when conducting qPCR experiments with new cohort of samples.

Algoritmo para el diagnóstico precoz de la deficiencia de vitamina b12 en ancianos

BACKGROUND: The elderly population is particularly at risk for developing vitamin B12-deficiency. Serum cobalamin does not necessarily reflect a normal B12 status. The determination of methylmalonic acid is not available in all laboratories. Issues of sensitivity for holotranscobalamin and the low specificity of total homocysteine limit their utility. The aim of the present study is to establish a diagnostic algorithm by using a combination of these markers in place of a single measurement. METHODS: We compared the diagnostic efficiency of these markers for detection of vitamin B12 deficiency in a population (n = 218) of institutionalized elderly (median age 80 years). Biochemical, haematological and morphological data were used to categorize people with or without vitamin B12 deficiency. RESULTS: In receiver operating curves characteristics for detection on vitamin B12 deficiency using single measurements, serum folate has the greatest area under the curve (0.87) and homocysteine the lowest (0.67). The best specificity was observed for erythrocyte folate and methylmalonic acid (100% for both) but their sensitivity was very low (17% and 53%, respectively). The highest sensitivity was observed for homocysteine (81%) and serum folate (74%). When we combined these markers, starting with serum and erythrocyte folate, followed by holotranscobalamin and ending by methylmalonic acid measurements, the overall sensitivity and specificity of the algorithm were 100% and 90%, respectively. CONCLUSION: The proposed algorithm, which combines erythrocyte folate, serum folate, holotranscobalamin and methylmalonic acid, but eliminate B12 and tHcy measurements, is a useful alternative for vitamin B12 deficiency screening in an elderly institutionalized cohort.

Use of Hy's law and a new composite algorithm to predict acute liver failure in patients with drug-induced liver injury.

BACKGROUND & AIMS Hy's Law, which states that hepatocellular drug-induced liver injury (DILI) with jaundice indicates a serious reaction, is used widely to determine risk for acute liver failure (ALF). We aimed to optimize the definition of Hy's Law and to develop a model for predicting ALF in patients with DILI. METHODS We collected data from 771 patients with DILI (805 episodes) from the Spanish DILI registry, from April 1994 through August 2012. We analyzed data collected at DILI recognition and at the time of peak levels of alanine aminotransferase (ALT) and total bilirubin (TBL). RESULTS Of the 771 patients with DILI, 32 developed ALF. Hepatocellular injury, female sex, high levels of TBL, and a high ratio of aspartate aminotransferase (AST):ALT were independent risk factors for ALF. We compared 3 ways to use Hy's Law to predict which patients would develop ALF; all included TBL greater than 2-fold the upper limit of normal (×ULN) and either ALT level greater than 3 × ULN, a ratio (R) value (ALT × ULN/alkaline phosphatase × ULN) of 5 or greater, or a new ratio (nR) value (ALT or AST, whichever produced the highest ×ULN/ alkaline phosphatase × ULN value) of 5 or greater. At recognition of DILI, the R- and nR-based models identified patients who developed ALF with 67% and 63% specificity, respectively, whereas use of only ALT level identified them with 44% specificity. However, the level of ALT and the nR model each identified patients who developed ALF with 90% sensitivity, whereas the R criteria identified them with 83% sensitivity. An equal number of patients who did and did not develop ALF had alkaline phosphatase levels greater than 2 × ULN. An algorithm based on AST level greater than 17.3 × ULN, TBL greater than 6.6 × ULN, and AST:ALT greater than 1.5 identified patients who developed ALF with 82% specificity and 80% sensitivity. CONCLUSIONS When applied at DILI recognition, the nR criteria for Hy's Law provides the best balance of sensitivity and specificity whereas our new composite algorithm provides additional specificity in predicting the ultimate development of ALF.

Identificació de subjectes a partir de l'anàlisi de dades multimodals RGB-Depth

En aquest treball s'explora el camp de la identificació facial de subjectes utilitzant tècniques d'anàlisi multimodal. Això és utilitzant imatges RGB i imatges de profunditat (3D) amb l'objecte de validar les diverses tècniques emprades en el reconeixement facial i aprofundir en sistemes que incorporen informació tridimensional als algorismes de detecció i identificació facial.