Asier Garmendia

Intraoral Digital Impressions for Virtual Occlusal Records: Section Quantity and Dimensions

The purpose of this study was to locate the 3D spatial position mandibular cast and determine its occlusal contacts in a novel way by using an intraoral scanner as part of the virtual occlusal record procedure. This study also analyzes the requirements in quantity and dimensions of the intraoral virtual occlusal record. The results showed that the best section combination consists of 2 lateral and frontal sections, the width of this section being that of 2 teeth (24 mm × 15 mm). This study concluded that this procedure was accurate enough to locate the mandibular cast on a virtual articulator. However, at least 2 sections of the virtual occlusal records were necessary, and the best results were obtained when the distance between these sections was maximum.

Virtual facebow technique.

This article describes a virtual technique for transferring the location of a digitized cast from the patient to a virtual articulator (virtual facebow transfer). Using a virtual procedure, the maxillary digital cast is transferred to a virtual articulator by means of reverse engineering devices. The following devices necessary to carry out this protocol are available in many contemporary practices: an intraoral scanner, a digital camera, and specific software. Results prove the viability of integrating different tools and software and of completely integrating this procedure into a dental digital workflow.

Predicting Patient Hospitalization after Emergency Readmission

ABSTRACT Emergency Departments (ED) suffer heavy overload due to lack of primary attention service. Increasingly geriatric admissions pose specific problems contributing to this overload. A consequence is the increase of patient returning short time after discharge, i.e., readmissions, sometimes requiring hospitalization. In this latter case the patient problem was not solved in the first admission and the condition has aggravated. The time threshold defining a patient comeback as readmission varies; therefore we have considered several such thresholds in our prediction experiments. Prediction of hospitalization following ED readmission is posed over a heavily imbalanced class distribution, so we have considered several approaches to deal with imbalanced datasets and several base classifiers, as well as performance measures that enhance the critical comparison between approaches. Experimental works are carried out on real data from a university hospital in Santiago, Chile, corresponding to a period of 3 years, including pediatric and adult admissions to the ED. We achieve results that encourage the development of real life application of the data balancing and classification approach for prediction of hospitalization after readmission.

Electrical Behavior Modeling of Solar Panels Using Extreme Learning Machines

Predicting the response of solar panels has a big potential impact on the economical viability of the insertion of alternative energy sources in our societies, diminishing the dependence on polluting fossil fuels. In this paper we approach the modeling of the electrical behavior of a commercial photovoltaic module Atersa A-55 using Extreme Learning Machines (ELMs). The training and validation data were extracted from the response of a real photovoltaic module installed at the Faculty of Engineering of Vitoria-Gasteiz (Basque Country University, Spain). The resulting predictive model has one input (\(V_{PV}\)) and one output (\(I_{PV}\)) variables. We achieve a Root Mean Squared Error (RMSE) of 0.026 in the electrical current measured in Amperes.

Making physical proofs of concept of reinforcement learning control in single robot hose transport task complete

Abstract This paper deals with the realization of physical proof of concept experiments in the paradigm of Linked Multi-Component Robotic Systems (LMCRS). The main objective is to demonstrate that the controllers learned through Reinforcement Learning (RL) algorithms with different state space formalizations and different spatial discretizations in a simulator are reliable in a real world configuration of the task of transporting a hose by a single robot. This one is a prototypical example of LMCRS task (extendable to much more complex tasks). We describe how the complete system has been designed and implemented. Two different previously learned RL controllers have been tested solving two different LMCRS control problems, using different state space modeling and discretization step in each case. The physical realizations validate previously published simulation based results, giving a strong argument in favor of the suitability of RL techniques to deal with LMCRS systems.

Towards Hospitalization After Readmission Risk Prediction Using ELMs

A criteria to evaluate the performance of Emergency Departments (ED) is the number of readmissions and hospitalizations short time after discharge of patients because the problem was not solved in the first admission. Such events contribute to overload the care system and to worsening the health of patients. In this paper we address the problem of predicting hospitalization events after readmission in ED, facing it as a classification problem and using Extreme Learning Machines (ELM). We have carried out experiments with a dataset with 45,089 admission events of 21,269 pediatric patients recorded in the Hospital Jose Joaquin Aguirre of the University of Chile during 3 years and 4 months, improving the state-of-the-art sensitivity results on the same dataset by 17%.

Neuronal Electrical Behavior Modeling of Solar Panels

In this paper authors model the electrical behavior of a commercial solar panel composed of solar cells connected in series through an Artificial Neural Network (ANN) with one hidden layer. The real solar panel that has been used as proof of concept is of the commercial model ATERSA A55, and it is placed at the Faculty of Engineering of Vitoria-Gasteiz (Basque Country University, Spain). The resulting model consists on one input (\(V_{PV}\)) and one output (\(I_{PV}\)), since the standard deviation of the temperature and irradiance magnitudes in the used dataset was residual.

Triage prediction in pediatric patients with respiratory problems

Abstract Respiratory diseases have an increasing prevalence in the large urban concentration of the world, due to the apparently unstoppable increase of air pollution from a diversity of sources. Children are specially a fragile part of the population suffering this conditions. Improved monitoring of critical patients by means of automatized data gathering and processing, i.e. alarm raising, aims to alleviate the risks of critical patients. Pediatric respiratory critical care has not received much attention in the literature, despite children care has specific conditions, such as the strong dependence of some physiological signals on the patient age. We approach the problem as triage prediction problem, formulated as multi-class classification problem, with special care on the age normalization of physiological variables. Data which can be used as classification features is scarce, in the sense that measurements of only a few variables are available, and that much of the qualitative information used by the medical doctors is not available. In this paper, we report the experimental results obtained on a data sample covering patients assisted in a local pediatric hospital during two years. The results conclude that it is possible to successfully predict the triage that the medical doctors will assign the critical patients. Success is mostly dependent on the features selected, specifically it is critical to include the triage in the previous record of the patient. That means that the caregivers follow very conservative decision policies. Besides, we have found that respiratory frequency is more discriminant than blood oxygen saturation.

Knowledge Modeling by ELM in RL for SRHT Problem

Single Robot Hose Transport (SRHT) is a limit case of Linked Multicomponent Robotic Systems (L-MCRS), when one robot moves the tip of a hose to a desired position, while the other hose extreme is attached to a source position. Reinforcement Learning (RL) algorithms have been applied to learn autonomously the robot control with success. However, RL algorithms produce large and intractable data structures. This paper addresses the problem by learning an Extreme Learning Machine (ELM) from the state-action value Q-table, obtaining very relevant data reduction. In this paper we evaluate empirically a classification strategy to formulate ELM learning to provide approximations to the Q-table, obtaining very promising results.

VALORACIÓN DE LOS ESTUDIANTES SOBRE LAS METODOLOGÍAS ACTIVA Y TRADICIONAL EN LA EUITI-BILBAO

This study analyzes the perception of engineering students of the EUITI of Bilbao on the achieved learning with active or traditional methodologies. It has been tested two samples of students: one composed of students who have previous experience and valued active teaching (PBL), and other sample who valued traditional teaching. No significant differences were observed between the two types of teaching in declarative knowledge. However, the development of essential skills in professional practice are much more developed using active methodologies. The reasons of preference for each methodology are also presented, and recommendations are offered in order to allow teachers develop strategies that promote in students a positive attitude towards learning, and facilitate the implementation of active teaching methodologies.