Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Advances in electrodermal activity processing with applications for mental health : from heuristic methods to convex optimization
Greco A., Valenza G., Scilingo E., Springer International Publishing, New York, NY, 2016. 138 pp. Type: Book (978-3-319467-04-7)
Date Reviewed: Oct 27 2017

Electrodermal activity (EDA) designates the variation in the electrical characteristics of the skin. The most pertinent property is skin conductance, quantified by measuring the current flow between two skin points when applying an electrical potential. EDA investigations have been around for more than 100 years, and meanwhile EDA has become a common measure of autonomic nervous system activity (ANS). Nowadays, EDA is connected to a large range of disciplines associated with emotional and cognitive states, arousal, or attention, and represents a robust indicator for emotional or psychopathological aspects. Recent improvements in technology have made EDA an increasingly important consideration in psychological science and in the new specific field of computer science, affective computing. The book presents an appraisal of the state of the art of EDA methodologies and applications, typical analysis tools, and recording means. An important point is the introduction of a new EDA model involving convex optimization. This method is investigated and evaluated in several phases of the presentation. The material takes advantage of the research carried out in the framework of two European projects, PSYCHE and WEARHAP. Thus some innovatory techniques and devices, and the corresponding results obtained during the unfolding of these projects, are presented throughout the material. The book has well-balanced and highly interdisciplinary content and might be of interest to a large audience, including psychologists, dermatologists, and physicians, or signal processing researchers, mathematicians, engineers, and scientists involved in affective computing.

The first chapter includes on the one hand an analysis of EDA and presents the underlying anatomical components and phenomena. On the other hand, it presents some EDA signal acquisition techniques. The authors emphasize several important aspects of EDA, useful to further understand modeling or acquisition issues such as the two important EDA components: the slow-varying tonic component, defined at skin conductance level (SCL), and the fast-changing phasic component, reflecting skin conductance response (SCR) to external stimuli. Subsequently, this chapter makes a general account of the existing EDA signal acquisition systems. The cornerstone of this part is the description of a textile wearable device able to perform EDA assessment based on both alternative (AC) and direct current (DC) injection. This seems a novel approach; most previous techniques only make use of DC, a few use AC, but none use both.

Chapter 2 reviews the existing EDA signal processing and modeling techniques. The authors evaluate and compare two methods. Conventional EDA modeling (CDA) is based on convolution between the two driving EDA components (SCL and SCR). CDA modeling involves deconvolution of the constituents and optimization through minimizing a certain cost function. The proposed convex optimization modeling (CvxEDA) brings in a number of new features, such as a probabilistic model with the acquired signal represented by additive tonic and phasic components, ARMA modeling, not the usual MA model, for the underlying IRF EDA component. A set of realistic premises allows further parameterization, maximum a posteriori probability, and underlying parameter estimation for each component by optimization of a resulting convex problem. The algorithm is implemented and available online. A last part of this chapter includes a description of the features, extracted in time and frequency domains, to be used in modeling.

Chapter 3 describes a set of tests undertaken for a first evaluation of the new CvxEDA algorithm. The procedure was tested on synthetic data and on two sets of real data from volunteers. Its performance was compared to CDA. An important goal of the testing session was to check the capacity of the new method to isolate the two main components, SCL and SCR, the ability to overcome SCR overlap when the time interval between consecutive stimuli is too short, and its power to distinguish among several types of arousal.

The fourth chapter presents some elements of the theory of emotions, surveys advances in ANS theory, and presents some ANS measures correlated to discrete emotional states, with direct application in bipolar disorder detection. Several definitions of emotions and modeling approaches are evoked. Methods for emotion induction in experimental or real clinical investigation and standardized tools for affective elicitation are also covered. An interesting section presents the state of the art in emotion recognition, today’s affective computing. A table accounting for very recent emotion recognition systems, based among other inputs on EDA signals, presents types of emotions investigated by each system, elicitation system used, performance achieved, and the corresponding bibliographical reference.

Chapter 5, the keystone of the material, presents the research results obtained applying EDA modeling to SC data, especially in detecting mood disorders and emotion recognition. A number of sections are dedicated to testing several research aspects when evaluating the two methods, conventional CDA and CvxEDA. For instance, apart from already standardized manners of inducing different levels of emotion by acoustic and visual stimuli, the book analyzes the affective response induced by more uncommon haptic (tactile) and olfactory stimuli. The performance is expressed through confusion matrices where percentages of true positives and true negatives reflect procedure efficiency.

By its structure and content, the book may serve as a guide to newcomers in psychological science and affective computing. It includes many interesting elements to guide future research activity, for example, a comprehensive bibliography including up-to-date scientific investigations. The book’s strength is its scientific description of the new EDA modeling technique, using interesting formal and graphical means, and the very-well-systematized experimental section that confirms the new method’s value.

The book begins with a list of acronyms, which is very useful. Unfortunately, not all of the abbreviations met throughout the book are contained in the list. For instance, ISI, SMNA, and LDC are missing. Including all terms--possibly in alphabetical order--would have been helpful. Other insignificant flaws might be in Figures 5.5 through 5.7 where top and bottom positions should be replaced by left and right.

Reviewer:  Svetlana Segarceanu Review #: CR145625 (1801-0008)
Bookmark and Share
 
Medical Information Systems (J.3 ... )
 
 
Convex Programming (G.1.6 ... )
 
 
Signal Analysis, Synthesis, And Processing (H.5.5 ... )
 
 
Optimization (G.1.6 )
 
 
Sound And Music Computing (H.5.5 )
 
Would you recommend this review?
yes
no
Other reviews under "Medical Information Systems": Date
The computer-based patient record
Dick R. (ed), Steen E. (ed), National Academy Press, Washington, DC, 1991. Type: Book (9780309044950)
Jan 1 1993
The doctor’s use of a computer in the consulting room: an analysis
Brownbridge G., Fitter M., Sime M. International Journal of Man-Machine Studies 21(1): 65-90, 1984. Type: Article
Mar 1 1985
Computing in medicine
Kendall ., Fagan L. (ed), Huang H., Pryor T. Computer 17(10): 294-313, 1984. Type: Article
May 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy