Traumatic brain injury (TBI) is a critical public health and socioeconomic problem throughout the world. Cognitive rehabilitation (CR) has become the treatment of choice for cognitive impairments after TBI. It consists of hierarchically organized tasks that require repetitive use of impaired cognitive functions. One important focus for CR professionals is the number of repetitions and the type of task performed throughout treatment leading to functional recovery. However, very little research is available that quantifies the amount and type of practice. The Neurorehabilitation Range (NRR) and the Sectorized and Annotated Plane (SAP) have been introduced as a means of identifying formal operational models in order to provide therapists with decision support information for assigning the most appropriate CR plan. In this paper we present a novel methodology based on combining SAP and NRR to solve what we call the Neurorehabilitation Range Maximal Regions (NRRMR) problem and to generate analytical and visual tools enabling the automatic identification of NRR. A new SAP representation is introduced and applied to overcome the drawbacks identified with existing methods. The results obtained show patterns of response to treatment that might lead to reconsideration of some of the current clinical hypotheses.
Today, it is well known that taking into account the semantic information available for categorical variables sensibly improves the meaningfulness of the final results of any analysis. The paper presents a generalization of mixed Gibert's metrics, which originally handled numerical and categorical variables, to include also semantic variables. Semantic variables are defined as categorical variables related to a reference ontology (ontologies are formal structures to model semantic relationships between the concepts of a certain domain). The superconcept-based distance (SCD) is introduced to compare semantic variables taking into account the information provided by the reference ontology. A benchmark shows the good performance of SCD with respect to other proposals, taken from the literature, to compare semantic features. Mixed Gibert's metrics is generalized incorporating SCD. Finally, two real applications based on touristic data show the impact of the generalized Gibert's metrics in clustering procedures and, in consequence, the impact of taking into account the reference ontology in clustering. The main conclusion is that the reference ontology, when available, can sensibly improve the meaningfulness of the final clusters.
Sànchez-Marrè, M.; Gibert, Karina; Vinayagam, R. K.; Sevilla-Villanueva, B. International Congress on Environmental Modelling & Software p. 493-500 Presentation's date: 2014-06-17 Presentation of work at congresses
Groundwater wells are one of the most important water resources in the world. Control and management of these resources are of high importance due to the implicit need of water as the main resource for life. This research focuses on a hydrogeological analysis with clustering, which is one of the most popular data mini ng methods, including In the classical data mining scheme, last step corresponds to the effective production of knowledge. In this paper, special focus on that part is done, by means of post - processing tools. The main goal is to discover prototypical profi les from the a c q uifer Pedro Gonzá lez in Marga rita Island (Venezuela), in order to understand the prototypical water conditions regarding quality and supply level. The database contain s 36 groundwater wells and their hydrogeological variables, i.e., electri c al conductivity, static level, pH and geographical coordinates that were collected in five annual measurement campaigns. Clustering methods were used to discover profiles and a typology of three types of wells was extracted. Post - processing tools were use d to get a conceptualization of the resulting classes and comprehensible profiles were finally described. The Class Panel Graph (CPG) and the Traffic Light Panel (TLP) were used to post - process the classes and understand the resulting profiles through symb olic visualization. The TLP was presented to the expert to support a multidisciplinary discussion and to create the mechanisms for a detailed understanding of the evolution of the aquifer. Results reported that the aquifer is in a critical situation in bot h water quality and supply levels. From this research, public administration performed some technical actions to improve the performance of the aquifer and its preservation. At present, predictive models local to profile s are developed
In real applications, important rates of missing data are often found and have to be pre-processed before the analysis. The literature for missing imputation is abundant. However, the most precise imputation methods require long time, and sometimes specic software; this implies a signicant delay to get nal results. The Mixed Intelligent-Multivariate Missing Im-
putation (MIMMI) method is proposed as a hybrid missing imputation methodology based on clustering. MIMMI is a non parametric method that combines the prior expert knowledge
with multivariate analysis without requiring assumptions on the probabilistic models of the variables (normality, exponentiality, etc). The proposed imputation values implicitly take into account the joint distribution of all variables and can be determined in a relatively short time. MIMMI uses the conditional mean according to the self-underlying structure of the dataset. It provides a good trade-o between accuracy and both simplicity and required time to data preparation. The mechanics of the method is illustrated with some case-studies, both synthetic and real applications related with human behavior. In both cases, acceptable quality results were obtained in short time.
Understanding the meaning of the classes outcomming from a clustering method is one of the criticalrn aspects to guarantee both the validity of the clustering results and their usefullness. The Methodology of conceptual characterization by embedded conditioning (CCEC), is a proposal for building conceptual interpretations of hierarchical clustering that contributes to enshort the gap between the clustering itself and the furtherrndecision-making processes. The methodologyrnuses some statistical tools (as the boxplot multiple,rnintroduced by Tukey,) together with some machine learningrnmethods, to learn the structure of the data; and find the characterizing variables (previously introduced by Gibert) of the classes when they exist, whereas providing alternatives when they do not exist. In this paper, the pillars of the methodology are presented, as well as different criteria for knowledge integration.rnThe usefulness of CCEC for building domain theories as models supportingrnlater decision-making is addressed. The proposal is applied for building therninterpretation of a set of classes extracted from a WasteWater Treatment Plant (WWTP) and the results obtained with the different criteria are compared and discussed.
Palomino, A.; Gibert, Karina International Conference of the Catalan Association for Artificial Intelligence p. 277-280 DOI: 10.3233/978-1-61499-452-7-277 Presentation's date: 2014 Presentation of work at congresses
Finding Internet browsing patterns is a current hot topic, with expected benefits in many areas, marketing and business intelligence among others. Discovering user's internet habits might improve fields like chained-publicity, e-commerce and media optimization. The large amount of data contained in log files that is currently being analyzed to find user's patterns require efficient and scalable data mining solutions. This paper proposes an algorithm to identify the most frequent route followed by Internet users, based on a specific combination of simple statistical and vectorial operators that provides exact solution with a really cheap computational cost. In the paper, the performance is compared with other two algorithms and an application to a real case study in the field of bussiness intelligence and chained publicity is presented.
Sevilla-Villanueva, B.; Gibert, Karina; Sànchez-Marrè, M. Frontiers in artificial intelligence and applications Vol. 256, p. 215-224 DOI: 10.3233/978-1-61499-320-9-215 Date of publication: 2013-11-01 Journal article
A profiling methodology is introduced for automatic interpretation of clusters in this work. This methodology contributes to the characterization of the resulting classes from a clustering process. This work aims to find a concordance between the proposed methodology and the experts’ description of these classes.
In this work the resulting classes from a clustering of a general population sample
based on their diet and physical activity habits are interpreted and compared with
the experts’ description of these classes by using the Class Panel Graphs. In this
work, we import techniques from the multivariate analysis into the cluster interpretation process.
Clustering techniques are commonly performed to find homogeneous and distinguishable prototypes. However, a careful interpretation of these prototypes is the key to assist the expert to better organize this knowledge for decision making support. We use the annotated Traffic Lights Panel (aTPL), not only as a postprocessing tool to help understand clustering prototypes, but also to manage uncertainty related to variability which is inherent by itself within the prototypes. The aTPL handles this uncertainty by using the variation coefficients (VC) in the classes over two dimensions - tone and saturation. Two different aTLPs were obtained to a WWTP in Slovenia. Results suggested that aTPL could be seen as a useful tool with good levels of reliability when interpreting and managing uncertainty related to decision-making based on clustering prototypes.
Traumatic brain injury (TBI) is the leading cause of death and disability in children and young adults around the world. There is not enough on-field experience yet regarding which specific intervention (tasks or exercises assignation) is more appropriated to help therapists to design their cognitive rehabilitation (CR) plans. Our proposal is to consider the CR treatment as a sequence of tasks and to determine the associations between a CR treatment (or relevant subsequences of it) and the degree of response of the patient to it. In the proposed methodology, a clustering process is performed in such a way that treatment profiles (classes) are identified.. Afterwards, responses to CR (improvements) of the patients placed in the different classes have been studied by means of conditional distributions of variables versus the classes. Analyzing CR tasks as treatment patterns offers a different perspective from the traditional single task focus, and may provide a comprehensive approach to therapists to design CR programs.
For almost twenty years the Catalan Association of Artificial Intelligence (ACIA) has been promoting cooperation between researchers in artificial intelligence within the Catalan speaking community.
This book presents the proceedings of the 16th International Conference (CCIA 2013), held at the University of Vic (UVIC), Catalonia, Spain, in October 2013. This annual conference aims to foster discussion of the latest developments in artificial intelligence within the community of Catalan countries, as well as amongst members of the AI community worldwide.
The book contains the 26 full papers, 5 short papers and 12 poster presentations from the conference, which are grouped under the following topics: relational learning, planning; satisfiability and constraints; perception and image processing; preprocessing; patterns extraction and learning; post-processing, model interpretability and decision support; recommenders, similarity and CBR; and multiagent systems.
Sevilla-Villanueva, B.; Gibert, Karina; Sànchez-Marrè, M. Conferencia de la Asociación Española para la Inteligencia Artificial p. 1454-1463 Presentation's date: 2013-09-18 Presentation of work at congresses
Nutritional Genomics studies diet-gene-disease interactions
and aims to promote health and disease prevention. It is based on the
idea that everything ingested into a person’s body affects the genome of
the individual and, therefore, both genes and nutrients modify the same
metabolic processes. This paper presents an application of clustering and
interpretation over real heterogeneous data coming from a nutritional
study. The individuals are clustered by their diet and physical activity
habits and the resulting clustering is interpreted. This work is part of a
methodology to deal with data from dietary intervention studies.
Gibert, Karina; Sànchez-Marrè, M.; Sevilla-Villanueva, B. Seminario Doctoral en Aplicaciones y Transferencia de la Inteligencia Computacional p. p01 Presentation's date: 2012-11 Presentation of work at congresses
Conti, D.; Gibert, Karina Congrés Internacional de l’Associació Catalana d'Intel·ligència Artificial p. 19-28 DOI: 10.3233/978-1-61499-139-7-19 Presentation's date: 2012-10-24 Presentation of work at congresses
In this work a proposal for making systematic state of the art is presented and applied to the Environmental Data Mining field. The main characteristics of the Data Mining process have been identified. A form has been created to check which of those characteristics take place in a real application and how. A random sample of Science Citation Index papers regarding Data Mining and Environmental Applications has been selected. Papers were read by a set of experts and a form was filled in for every paper. The resulting information was mined itself using basic statistical analysis and some specific treatments for multi-response variables, to get a first picture of what is currently being done in the applications of Data Mining methods to environmental fields. Very interesting results have been obtained which depict very useful information. This information ranges from a general picture of what kind of methods are commonly used to which environmental areas seems to be more deeply using the data mining techniques.
The paper presents and discuss these results, together with a proposal for building a continuous collaborative pannel in the web for enlarging the sample
of papers and update the picture continuously. This will be easily possible because
the analysis of the recorded data has been automatized in a statistical package
set of macros for repetitive updating mined knowledge. The proposal is oriented to provide an Environmental Data Mining Observatoire, where getting updated
information on what is being done in the area, identify drawbacks, orient future
research in the methodological field to provide answer to the open environmental
problems and finally, to give the environmental audience a wide corpus of previous experiences to be used as a reference for new applications
This paper describes the integral Knowle dge Discovery (KDD) process, including bot h prior expert knowledge and interpretation oriented tools to extract the behavior of a real pilot wastewater treatment plant. Special emphasis is made on the interest of developing postprocessing tools for clustering methods which can help the expert to unde rstand the meaning of the clusters and bridge the important existing gap between Data Mining and effective Decision Support. Traffic Lights Panel (TLP) is presented a s a suitable visual interpretation oriented tool for clustering results. Based on this tool, four typical behaviours are identifi ed in the pilot plant, which have been validated by the experts. Till now, the TLP is manually derived from the clustering results, but i t has been well accepted by the domain experts of several real applications as a very helpful contribution to understand the classes meaning and improve reliable decision-maki ng. Here, a proposal for automatic construc tion of TLP is presented trying to mimic the real process that the analyst perform s to manually build them. A criterion based on conditional Median as a central trend statistics of the variables inside a class is introduced and re fined to gain robustness towards outliers. Both criteria are tes ted and compared with the real target case study. A deep analysis of the advantages and draw backs of the proposed criterion, permitted to better understand the analyst process when manually building TLPs , to identify the scope of the proposal, and to typify some of the situations in which additional conditions are required.
Batet, M.; Isern, D.; Marín, L.; Martínez, S.; Moreno, A.; Sánchez, D.; Valls, A.; Gibert, Karina Journal of intelligent information systems Vol. 38, num. 1, p. 95-130 DOI: 10.1007/s10844-010-0145-0 Date of publication: 2011-12-06 Journal article
Home Care (HC) assistance is emerging as an effective and efficient
alternative to institutionalized care, especially for the case of senior patients that present multiple co-morbidities and require life long treatments under continuous supervision. The care of such patients requires the definition of specially tailored treatments and their delivery involves the coordination of a team of professionals from different institutions, requiring the management of many kinds of knowledge (medical, organizational, social and procedural). The K4Care project aims to assist the HC of elderly patients by proposing a standard HC model and implementing it in a knowledge-driven e-health platform aimed to support the provision of HC services.
The SUPERHUB project aims at realizing a new services mobility framework supporting an integrated and eco-efficient use of multi-modal mobility systems in an urban setting.\nSUPERHUB provides a user-centric, integrated approach to multi-modal smart urban mobility systems, through an open platform able to consider in real time various mobility offers and provide a set of mobility services able to address user needs, promote user participation and to foster environmental friendly and energy–efficient behavioural changes. Moreover, the take-up of virtuous behaviours, characterized by a reduced environmental footprint, is also facilitated by the SUPERHUB open platform matchmaking and negotiation capabilities between (public-private) providers and consumers of mobility offers and by the use of persuasive technologies to achieve wide adoption of results.\nTo achieve these objectives SUPERHUB will develop a persuasive engine based on captology principles to facilitate the voluntary adoption of environmentally-friendly multi-mobility habits, novel methods and tools for real-time reasoning on large data streams coming from heterogeneous sources, new algorithms and protocols for inferring traffic conditions from mobile users by coupling data from mobile operator networks with information coming from GPS based mobile phones and for dynamic matchmaking or resources that will generate journey plans best fulfilling user mobility needs and preferences while minimizing negative environmental impact.\nSpecific services and user studies will be realized to demonstrate the SUPERHUB concepts and technologies through field trials in Barcelona, Helsinki and Milan, involving large end-users communities. Results of SUPERHUB field trials shall influence local policy makers and municipalities in the definition of new energy-aware mobility strategies and planning.