Ortuño Guzmán Francisco M. - Bioinformatics and Biomedical Engineering: 5th International Work-Conference, IWBBIO 2017, Granada, Spain, April 26-28, 2017, Proceedings

Bioinformatics and Biomedical Engineering: 5th International Work-Conference, IWBBIO 2017, Granada, Spain, April 26-28, 2017, Proceedings — read online for free the complete book (whole text) full work

Below is the text of the book, divided by pages. System saving the place of the last page read, allows you to conveniently read the book "Bioinformatics and Biomedical Engineering: 5th International Work-Conference, IWBBIO 2017, Granada, Spain, April 26-28, 2017, Proceedings" online for free, without having to search again every time where you left off. Put a bookmark, and you can go to the page where you finished reading at any time.

Light

Font size:

↓

↑

Georgia Tahoma Arial Verdana

Reset

Interval:

↓

↑

Bookmark:

Make

1234567...218

Advances in Computational Intelligence for Critical Care

Springer International Publishing AG 2017

Ignacio Rojas and Francisco Ortuo (eds.) Bioinformatics and Biomedical Engineering Lecture Notes in Computer Science 10208 10.1007/978-3-319-56148-6_1

Health Monitoring System Based on Parallel-APPROX SVM

Fahmi Ben Rejab 1

(1)

BESTMOD, Institut Suprieur de Gestion de Tunis, Universit de Tunis, 41 Avenue de la Libert, 2000 Le Bardo, Tunisie

Fahmi Ben Rejab (Corresponding author)

Email:

Walid Ksia

Email:

Kaouther Nouira

Email:

Abstract

In this paper, we propose a new approach to deal with the high rate of false alarms generated by the health monitoring system (HMS) in intensive care units (ICU). We propose a new HMS based on a new classification method consisting of the parallel-Approx support vector machines (PASVM). The main aim of the new HMS denoted by PASVM-MS is to considerably reduce the rate of false alarms and to make a fast prediction in each new state of the patient. Besides, it overcomes the main issue of the existing HMS by proposing a classification model that considers the variation of the patient states over time. Also, the number of measured parameters have to be changed when patients are getting better by removing one or more variable each time. However, thresholds are stable and do not translate the states of patients over time, since all existing systems in ICU do not take into account of the patients states evolution. Our proposal is able to generate an initial model classifying states of patients to normal and abnormal (critical) using the PASVM. Then, it updates its model by considering the evolution in the states of patients using PASVM especially when we deleting one variable. As a result, the new system gives what the medical staff wants as information and alarms relative to monitored patient.

Keywords

Health monitoring system Parallel SVM Approx SVM False alarms Intensive care unit

Introduction

Intensive care unit (ICU) is a special department in hospital devoted to patients whose conditions are life threating. It provides intensive care to patients and try to control critical states. ICU is equipped by sophisticated monitoring devices such as health monitoring system (HMS). The principal aim of HMS is to measure and alert medical staff when patient has a critical state. The current HMS function is based on threshold set by care-givers. An alarm is trigged when its limits are violated.

In many research studies [] authors have reported that in Massachusetts General Hospital in 2010 the result of turning off alarms have caused the death of a patient.

Improving the HMS in ICU, reducing the number of false alarms, and increasing the positive ones have attracted the intention of many researchers over the past ten years. Several researchers have focused on this problem in ICU. We can mention the use of the digital signal processing in [] that have reported and detailed this issue and how to overcome it.

To this end, we propose in this paper a new HMS based on a new machine learning technique which is the parallel and Approx support vector machines (PASVM). The PASVM is a modified version of the standard SVM which can improve the model of classification in the test phase when we decrease the number of attributes. Our proposal reduces the number of ignored or ineffective alarms, makes a faster prediction in the patients state, and takes into account the evolution of his state as well as the number of measured parameters over time.

The rest of the paper is structured as follows: Sect. concludes the paper.

Health Monitoring System in ICU

Monitoring patient in critical care environments such as intensive care units (ICUs) and operating rooms involves estimating the status of the patient, reacting to events that may be life-threatening, and taking actions to bring the patient to a desired state. There is the use of medical devices when monitoring patients. Besides, the current HMS generates important data and information relative to the monitored patient. Each measured variable has a practical limits or threshold, when a parameter exceeds its limits an alarm is trigged.

Table shows an example of some measured parameters with their thresholds.

Table 1.

Some measured parameters

Medical parameters	Minimum value	Maximum value
Heart rate
Respiratory rate
Pulse rate
Saturated percentage of Oxygen in the blood

Alarms are generated by crossing a given limit. Unfortunately, it is not the best method to indicate the patient states. There is not a consideration of the simultaneous evolution of different parameters. The information that the medical staff generally wants is the detection of critical changes in a patient conditions.

Studies have demonstrated that the majority of alarms created by patient monitoring systems have no clinical relevance. Borowski et al. [], only 10% of all alarms were taken into account by care-givers and 50% of all relevant alarms were not correctly identified.

With the large number of false alarms, care-givers do not trust the used monitoring system anymore and are becoming desensitized. As a results, care-givers ignore the majority of alarms and consider this system as a measurement tool not as a monitoring one. Due to the high rate of false alarms, the sensitivity of the current HMS is not close to 100%. There were 75 life-threatening situations, where no alarm occurred, reported in the Federal Institute for Drugs and Medical Devices (BfArM) in Germany. Besides, missing true alarms caused, between 2002 and 2004, 237 deaths related to device alarms [].

As a result, the future patients monitoring system has to allow the medical staff to be more confident in the HMS. It has not to be a simple measurement tool but a monitoring one by providing important medical information. In order to avoid the issues indicated above, the monitoring system should be improved. Hence, we propose a new HMS based on parallel and Approx support vector machines. This latter makes it possible to detect different patients states as desired and needed by medical staff. The PASVM is trained from expert decisions and can simulate the expert task for new observations especially when we decrease the number of measured parameters.

Parallel and Approx Support Vector Machines

3.1 The Standard SVM

Support vector machines (SVM) has been introduced by Vapnik in 1995 []. The basic idea was to find an hyperplane which separates data into its two classes with a maximization of the margin.

Fig. 1.

The optimal hyperplane

1. Case of linearly separable data: Given a training set of observations where

Next page

1234567...218

Light

Font size:

↓

↑

Georgia Tahoma Arial Verdana

Reset

Interval:

↓

↑

Bookmark:

Make

Similar books «Bioinformatics and Biomedical Engineering: 5th International Work-Conference, IWBBIO 2017, Granada, Spain, April 26-28, 2017, Proceedings»

Look at similar books to Bioinformatics and Biomedical Engineering: 5th International Work-Conference, IWBBIO 2017, Granada, Spain, April 26-28, 2017, Proceedings. We have selected literature similar in name and meaning in the hope of providing readers with more options to find new, interesting, not yet read works.

Qi Liu

Proceedings of the 11th International Conference on Computer Engineering and Networks

Sara Paiva

Science and Technologies for Smart Cities: 6th EAI International Conference, SmartCity360°, Virtual Event, December 2-4, 2020, Proceedings

Rick Capozzi

The Growth Mindset

Pietrusiak Damian

Proceedings of the 13th International Scientific Conference: Computer Aided Engineering

Korytkowski Marcin.

Artificial Intelligence and Soft Computing: 16th International Conference, ICAISC 2017, Zakopane, Poland, June 11-15, 2017, Proceedings, Part I

Grünbacher Paul

Requirements Engineering: Foundation for Software Quality: 23rd International Working Conference, REFSQ 2017, Essen, Germany, February 27 - March 2, 2017, Proceedings

Baumeister Hubert

Agile processes in software engineering and extreme programming: 18th International Conference, XP 2017, Cologne, Germany, May 22-26, 2017, proceedings

Corchado Emilio

International Joint Conference SOCO16-CISIS16-ICEUTE16: San Sebastián, Spain, October 19th-21st, 2016 Proceedings

Amsaleg Laurent

MultiMedia Modeling: 23rd International Conference, MMM 2017, Reykjavik, Iceland, January 4-6, 2017, Proceedings, Part II

Amann Johanna

Passive and Active Measurement: 18th International Conference, PAM 2017, Sydney, NSW, Australia, March 30-31, 2017, Proceedings

Alves José C.

Robotic Sailing 2016 Proceedings of the 9th International Robotic Sailing Conference

Domenico Salvagnin

Integration of AI and OR techniques in constraint programming: 14th international conference, CPAIOR 2017, Padua, Italy, June 5-8, 2017: proceedings

Discussion, reviews of the book Bioinformatics and Biomedical Engineering: 5th International Work-Conference, IWBBIO 2017, Granada, Spain, April 26-28, 2017, Proceedings and just readers' own opinions. Leave your comments, write what you think about the work, its meaning or the main characters. Specify what exactly you liked and what you didn't like, and why you think so.