TECNIO
eXiT- Control Engineering and Intelligent Systems is a TECNIO accredited agent.
The TECNIO certification created by the Government of Catalonia, through ACCIÓ, identifies differential applied technology providers and facilitators.
Collaborations (most recent)
AI FORA – Artificial Intelligence for Assessment
Volkwagen.
Albert Sabater, Beatriz Lopez. 2020.
Dreamer: Detección precoz alzheimer
MJN Neuroserveis.
Beatriz Lopez, Jaume Gauchola, Joaquim Massana. 2020-2023.
Algorithms to predict epileptic seizures.
MJN Neuroserveis.
Beatriz Lopez. 2019.
Testing and analysis of artificial intelligence models that can anticipate epileptic seizures minutes in advance using clinical data from real patients.
MJN Neuroserveis.
Beatriz Lopez. 2017-2018
Establish a framework for action with regard to the collaboration between the Parties in the execution of a real-life pilot test.
FMWOR – Fundació Barcelona Mobile World Capital Fundation MJNNE – MJN Neuroserveis S.L. 4919 – Clínica Corachán REESP – Rebar Espai SL (Matsalud).
Beatriz Lopez. 2017-2018.
Dissemination of the HTE-DLP 3.0 tool, which aims to improve the management of Familial Hypercholesterolaemia, in specialised scientific and technical forums throughout 2018.
SANOFI
Alberto Zamora, Beatriz Lopez. 2018
Prediction of the number of emergency visits and the need for emergency hospitalisation.
FHPA – Fundació Hospital de palamós (FUNDACIÓN MN. MIQUEL COSTA)
Beatriz Lopez. 2015-2018.
Intelligent management of drinking water distribution networks based on online and offline data supplied by PRODAISA.
PRODAISA.
Joan Colomer. 2020-2021.
The study of flexibility that charging sessions of electrical vehicles can provide and the elaboration of an optimal charging strategy for a more efficient use of electricity.
Resourcefully.
Joaquim Melendez, Marc Cañigueral. 2019-2021
Study of the possibilities offered by the data mining tools available to the eXiT group in the field of management of drinking water distribution networks.
PRODAISA
Joan Colomer. 2017-2018.
Study of the possibilities offered by the data mining tools available to the eXiT group in the field of energy management, in the case of buildings managed by AITEL.
AITEL
Joan Colomer. 2017-2018.
Advanced systems for process management of RSU plants
DNVG – Det Norske Veritas Business Assurance España S.L.
Joan Colomer Llinas. 2015-2016.
Tools
Case Based Reasoning (CBR) algorithms for decision support. Case-based Reasoning has been demonstrated to be a powerful methodology to build recommended systems, allowing to define individual and adaptive recommendations. Case-based reasoning consists of four main stages, retrieve, reuse, revise and retain, that could be extended with additional ones regarding maintenance, as review and restore. In each stage different techniques can be selected according to the application domain (e.g. Euclidean or hamming distance in retrieve, voting or aggregation in reuse, simulation in revise, always retain).
Two important modes of working: on-line to tune the hyper-parameters of a CBR system, and a library than can be generated that includes the final outcome of the tuning phase, so as to integrate the CBR in any run-time environment.
This tool has been used previously in the Pepper H2020 project to build an adaptive insulin recommender system, in the ITEA Moshca project to monitor premature babies at home, and some other older projects inside the healthcare domain (migraines, cancer, …) as well as others (moulding, energy). In each project, new methods and libraries are begin added according to the particularities of the problem to be solved. For example, for the insulin recommender system, a method for reusing past solutions has been added.
Java.
Cite as: Beatriz López, Carles Pous, Albert Pla, Pablo Gay, Judith Sanz, and Joan Brunet. eXiT*CBR: A framework for case-based medical diagnosis development and experimentation. Artificial Intelligence in Medicine. 51, 2011, 81-91.
Some variants of this tool includes:
- exitCDDS: for workflow management. Java. Cite as: Andres El-Fakdi, Francisco Gamero, Joaquím Meléndez, Vincent Auffret, Pascal Haigron. eXiTCDSS: A framework for a workflow-based CBR for interventional Clinical Decision Support Systems and its application to TAVI. Expert Syst. Appl. 41(2): 284-294 (2014)
- Ceaseless CBR, for sequence of events. Python. Cite as: publication under review.
- Context-aware reasoning. Java. Cite as: Albert Pla, Jordi Coll, Natalia Mordvaniuk and Beatriz López. Context-Aware Case-Based Reasoning. The Second International Conference on Mining Intelligence and Knowledge Exploration (MIKE), Cork, 2014. LNAI 8891, pp. 229-238, 2014.
Library to facilitate the creation of systems of rule-based systems for monitoring or prediction purposes. The goal is to facilitate the final user (experts) the introduction of the rules.
This tool has been previously used in the ITEA Moshca project to monitor premature babies at home, and in the national funding project TDApp (Personalized treatment of ADHD) to implement the Cochare/grade decision making procedure for intervention recommendations.
Java
Cite as: B. López, J. Coll, F.I. Gamero, E. Bargalló, A. López-Bermejo. Intelligent systems for supporting premature babies healthcare with mobile devices. Mobilemed 2013, 4 pages
Several algorithms related to sequence learning pattern mining considering different types of patterns to be learn: sequences, with informed gaps, multiset, temporal relations.
The algorithms have been applied to different smart cities problems, as well as in manufacturing (complex event systems).
Python, Java.
Cite as:
Pablo Gay, Beatriz López and Joaquim Meléndez. Learning Complex Events from Sequences with Informed Gaps. IEEE 14th International Conference on Machine Learning and Applications (ICMLA), Miami, FL, USA, 2015, pp. 1089-1094.
Pablo Gay, Beatriz López and Joaquim Meléndez, Constraint-Programming Approach for Multiset and Sequence Mining. In Ana Fred and Joaquim Filipe (Eds.): Proceedings of the 4th International Conference on Knowledge Discovery and Information Retrieval (KDIR), 4rd to 7th October 2012, Barcelona, Spain, pages 212-220. ISBN: 978-989-8565-29-7.
Algorithm to mine temporal patterns (i.e. A before B) from time series.
Python. Our implementation of vertTIRP is available from the Bitbucket repository at ‘‘https://bitbucket.org/natalia_mordvanyuk/verttirp/src/master/’’. The following username and password can be used to log in:
username: invitedbymordvanyuk@gmail.com
password: YgGhjhGllgFFdlkbvFM..,
Cite as:
Natalia Mordvanyuk, Beatriz López, Albert Bifet: vertTIRP: Robust and Efficient Vertical Frequent Time Interval-related Pattern Mining, Expert Systems with Applications, Volume 168, 15 April 2021, 114-276.
Signal eeg
Signal data processing tool kid, with special emphasis on wearable data and EEG signals. Functionalities for all steps of data mining: noise filtering, windowing, feature extraction, feature selection, modelling, data balancing, validation methods, visualisation.
This toolkit has been used for the national funding project SERAS, for seizure detection, as well as applied for alcoholism prognosis, schizophrenia, and emotion recognition.
In the Repair project it is expected to use the tool for managing data from wearables. Particular modifications could be considered for their integration as a library.
Matlab. Available at: https://caleta.udg.edu/git/eXiT_Research_Group/Signaleeg
Cite as:
Joaquim Massana, Òscar Raya, Jaume Gauchola, Beatriz López: Signaleeg. A practical tool for EEG signal data mining, Neuroinformatics, Neuroinformatics volume 19, pages 567–583 (2021).
Products
Adaptive Insulin recommender. Based on exitCBR.
Android
Cite as: Ferran Torrent-Fontbona, Beatriz Lopez. Personalised Adaptive CBR Bolus Recommender System for Type 1 Diabetes. IEEE Journal of Biomedical and Health Informatics , Vol 23, Num. 1, January 2019, pp. 387-394
Platform for monitoring premature babies at home.
Second prize at the VdH Innovation Days 2015.
Java.
Cite as: Natalia Mordvanyuk, Beatriz López, Montserrat Reixach, Marta Fabrellas, Montserrat Planella, Carles Cordon, Nuria Simon, Anna Duran, Josep Perapoch, Judit Bassols, Abel López-Bermejo: NoaH: Supporting Premature Babies Care with Mobile Phones, MSF Peadriatics Days, Stockholm, Sweden, April 5-6, 2019.
Decision support tool for treatment recommendation for dyslipidaemia and familial hypercholesterolemia.
A. Zamora, B. López and F. Torrent have the intellectual property register of HTE 3.0.
Java.
Cite as: Beatriz López, Ferran Torrent-Fontbona, Guillem Paluzie, Alberto Zamora. HTE 3.0: Clinical Decision Support System for lipid-lowering treatment and familial hypercholesterolemia detection. 8enes Jornades de TIC Salut i Social, Vic, Catalonia, September 27-28 2018.
Gait analysis for prediction rehabilitation length after a hip surgery. Based on Bag of words for feature extraction and SVM and Case-Based-Reasoning for classification.
Java
Cite as: Albert Pla, Natalia Mordvanyuk, Beatriz López, Marco Raaben, Taco J. Blokhuid, Herman R. Holstlag: Bag-of-steps: Predicting Lower-limb Fracture Rehabilitation Length. Neurocomputing, Volume 268, 13 December 2017, Pages 109-115.
The Data Pre-Processing Application is a tool that detects/corrects missing, corrupt or inaccurate (outliers) data, re-samples them, if needed, and gets energy load profiles (daily, weekly). This is a necessity to exploit these information by other tools (forecasting, optimization, planning).
Download the Product Sheet (in E-LAND project)
The Optimal Scheduler tool provides an hourly scheduling of storage (when store or consume) and controllable assets (when switch on/off) in order to maximize the use of renewable energy resources. It is based on the forecast production/consumption in the Local Energy System (LES). The application is fully integrable in the Energy Management System.
The Optimal Scheduler module provides the scheduling of operating points of available storage units and flexible loads (that can be rescheduled) that optimize the use of local renewable energy sources. Multi-vector energy is considered, including vector energy transformations.
A suite of web services that provides enhanced energy monitoring and scheduling capabilities.
PROBLEM ADDRESSED: The presence of RES in the grid requires additional LV grid observability and support mechanisms for more efficient management. Traditional methods to overcome these challenges require large investments.
VALUE PROPOSITION:The LVD-DST provides DSOs with the intelligence to convert their LV
grid into a Smart LV grid.
KEY FUNCTIONALITIES:
- Demand & generation forecasting: uses smart meter data for day-ahead forecasts and prediction of congestion and voltage
problems. - Fault detection & isolation: provides enhanced monitoring of the grid based on multivariate statistics to automatically detect faults
and other abnormalities in a statistical sense. - Optimal grid operation scheduling: calculates optimal grid operation schedules of active elements (switchgear/storage) in
the grid to prevent critical events, reduce energy exchange at substation level and peak shaving.
Renewable sources of energy, such Photovoltaic or wind generation, are intermittent. In order to maximize the use of the energy they can generate, consumed or stored, it is needed to estimate the expected generation and the expected consumption. The Energy Forecaster provides the tool to forecast both generation and consumption.
- The tool provides forecasting for different energy vectors: electrical and thermal loads; Photovoltaic and wind generation.
- Data provided to the Energy Forecaster tool are first pre-processed by the Data Pre-Processing Application tool in order to ensure their quality.
- Forecasting results are provided to the Optimal Scheduler tool in order to calculate the optimal scheduling of assets.
The Energy Forecaster tool provides hourly forecasting of electrical/loads and Photovoltaic/wind generation. Two forecasting horizons are provided: intra-day and day-ahead. Forecasts are based on weather data, characteristics of generation assets, and contextual information. Occupancy also can be considered as an input for forecasting. The application is fully integrable in the Energy Management System.
Download the Product Sheet (in E-LAND project)
Spin off- Newronia
NEWRONIA is a Technologic Based Enterprise founded in 2011 as a result of the activity in the eXiT research group of Universitat de Girona to solve optimisation problems.
- Artificial intelligence–> Develops computer applications based on artificial intelligence that help in the decision making, specially when a large number of variables and constraints are involved.
- Optimization–> Carries out project to optimize the use of resources in enterprises, achieving a reduction in costs,increase productivity andminimizing environmental impact
- Distribution and transport–> The daily process of deciding the route for each vehicle is very repetitive and tedious and it can be automated, saving time and finding solutions that minimize costs, among other criteria.
