Author: Mr Hugo Marques STSM Period: 2021-07-12 – 2021-07-26
Hosting institution: Marche Polytechnic University
PURPOSE OF THE STMS
The proposed STSM aims to develop algorithms able to process and elaborate physiological measurements collected from subjects by the use of wrist-worn devices, within their workplace environment. It is in fact well-known that the stimuli originating from the surrounding environment may have a different impact on different subjects, with strong inter- and intra-subject variability. This study aims to look for quantitative footprints of different reactions generated on a subject by analyzing their Galvanic Skin Response (GSR), also named ElectroDermal Activity (EDA), measurement data when exposed to acoustic stimuli. In particular, it is of interest to investigate the properties of the GSR signal, and how they change in response to the stimuli provided. The results of the STSM are expected to contribute to the activities of the subgroup WG4.3 on ICT solutions for ageing well in the workplace, as it will identify objective metrics or features to look for when sensing the subject’s status by means of wearable devices.
DESCRIPTION OF WORK CARRIED OUT DURING THE STSMS
During my research visit, the collaboration with Dr. Susanna Spinsante and her team was organized in three main tasks.
The following tasks were performed in the proposed STSM:
T1: State-of-the-art research, by reading the scientific and technical literature about the known properties and behavior of the GSR signal, with respect to different external stimuli, with a focus on acoustic stimulation. Also, I read and researched all the documentation about the Ledalab (http://www.ledalab.de/) software to understand how it works and the functionalities that could be useful (for example: optimizing the data with «The Continuous Decomposition Analysis (CDA)» of Ledalab).
T2: Data collection, by participating in experimental tests performed with some volunteers to be monitored through the GSR sensor (EDA) during rest and sound stimulation. In particular, 10 subjects (5 males and 5 females, aged between 20 and 33 years) were included in the study to investigate the physiological response to different sound stimuli. When collecting data, we divided them into two phases of data
collection (rest phase and stimulus phase).
For this data collection we used simultaneously the ProComp Infinity (laboratory instrument) and the Empatica E4 device (consumer wearable device), and we focused first on the processing of Empatica E4 data used later in the Matlab and data analysis step.
T3: Implement proper algorithm (in Matlab environment) to pre-process GSR data and then extract features (using Ledalab tool), by quantifying the effects of acoustic stimulation on the subjects. Matlab was used to separate the data, using the «tags» file that corresponds to a sequential list of values in seconds corresponding to the instants when the button was pressed, according to the type of sound stimulus received and in relation to the state of rest of the person, which allows dividing and identifying the part with stimulus and without stimulus.
I analyzed information previously obtained, through its optimization and extraction to excel documents, respectively organized, as well as graphically in Matlab in order to detect a variation of Galvanic Skin Response values in each of the different sound stimuli applied to each person, and if possible to detect a pattern of behavior of values, in an overview.
DESCRIPTION OF THE MAIN RESULTS OBTAINED
This chapter describes the results obtained during the STSM and also presents respective explanations for clarification.
With the separation of data into two parts to differentiate which is the portion stimulated or not, I used the code , that I exported to both types of files (.csv & .txt) because in this case Ledalab only accepts importing in text format, and in that sense, it was useful to have both final formats.
And with the help of Ledalab, an online available tool implemented in Matlab, by applying the Continuous Decomposition Analysis (CDA) it enabled the optimization of two parameters (i.e. tau1 and tau2) that define the signal shape, according to the model of a biexponential function expressed in the literature.
With all this processing and data extraction, it was possible to easily observe that most people suffered changes in their Galvanic Skin Response values when they were under the effect of a sound stimulus and depending on the sound stimulus there were different value peaks. From another perspective, using Ledalab also enabled the easier analysis of these peaks with more detailed graphs when optimized.
Also, a comparison was made even more in depth, dividing the data with stimuli in the 4 parts (eq1 to eq4) with the different 4 sound stimuli in order to analyze the number of peaks in each part, and understand what conclusions we could reach.
The following results are expected from this proposed STSM:
R1: The data collected from experiments and analysed by me will be checked for a joint publication by the researchers in Polytechnic Institute of Viseu and UNIVPM on the topic of the STSM, hopefully by the end of the year. R2: Following the 2 weeks STSM that allowed me to set up all the tools necessary to process the data, we will continue collaborating, by means of remote collaboration tools and online tools, to work towards the publication and possibly the processing of new data collected in future acquisitions.