The Alexithymia Database consists of two datasets that include five physiological signals and several digital channels associated with the experiment state. It is designed to provide insights into the emotios of people with alexithymia. This readme file aims to guide users on how to effectively use and understand the datasets.
The dataset files are provided in two formats: ACQ files and a comma-separated values (CSV) format file for demographics information.
- ACQ files: The datasets are stored in ACQ files, which are data graph files created by AcqKnowledge. Each ACQ file represents data for a single participant. These files contain the recorded physiological signals and various digital channels related to the state of the experiment.
- Demographics CSV file: In addition to the ACQ files, a separate CSV file is available to access demographics information about the participants. This file provides details such as ID, age, gender, group label, etc. It complements the ACQ files by providing additional contextual information about the participants.
Channel ECG: IgnoreChannel SCR: Skin conductance response levelChannel ORB: IgnoreChannel COR: IgnoreChannel ZYG: IgnoreChannel Sound: IgnoreChannel phase1: Digital channel indicating first imagery periodChannel Arousal: Digital channel indicating low/high arousal of emotionChannel Valence1: IgnoreChannel Valence2: Digital channel indicating the presence of negative emotionChannel Probe ON: IgnoreChannel Tone1: Digital channel indicating shallow depth of processingChannel Tone2: Digital channel indicating deep depth of processingChannel phase2: Digital channel indicating second imagery periodChannel ORB Filter: IgnoreChannel COR Filter: IgnoreChannel ZYG Filter: IgnoreChannel ECG BPM: ElectrocardiogramChannel Trial counter: IgnoreChannel ORB Int: Electromyography of orbicularisChannel COR Int: Electromyography of corrugatorChannel ZYG Int: Electromyography of zygomaticusChannel ECG R-R: IgnoreChannel Pulse Rate: Ignore
We are interested into specific columns, therefore follow the instructions below to extract only the necessary data.
The relationship between arousal and valence can help interpret emotional states in the dataset. Based on the provided mappings, the following associations can be made:
- When Arousal = 5 and Valence2 = 0, it corresponds to the emotion of Joy.
- When Arousal = 0 and Valence2 = 0, it corresponds to a Neutral emotional state.
- When Arousal = 5 and Valence2 = 5, it corresponds to the emotion of Fear.
These mappings can provide insights into the emotional experiences represented in the dataset.
The data collection protocol consisted of three distinct phases: a 20-second baseline, a 60-second phase 1, and a 40-second phase 2. Each participant completed a total of 10 trials, resulting in a cumulative data duration of 1200 seconds for each participant.
To identify and extract the relevant data for analysis, it is important to focus on the specific digital channels corresponding to the Baseline, Phase1, and Phase2 periods. The data collected between these phases should be disregarded as it does not pertain to the targeted experimental periods.
By filtering the dataset using the digital channels associated with the Baseline, Phase1, and Phase2, you can ensure that only the desired segments of data are considered for further analysis.
- When Phase1 = 0 and Phase2 = 0, it corresponds to the phase of Baseline.
- When Phase1 = 5 and Phase2 = 0, it corresponds to the phase of Phase1.
- When Phase1 = 0 and Phase2 = 5, it corresponds to the phase of Phase2.
These mappings can provide insights into the phase of the experiment in the dataset.
By filtering the dataset using the digital channels associated with the Tone1 and Tone2, you can ensure that only the desired segments of data are considered for further analysis.
- When Tone1 = 5 and Tone2 = 0, it corresponds to the phase of Shallow.
- When Tone1 = 0 and Tone2 = 5, it corresponds to the phase of Deep.
These mappings can provide insights into the depth of processing of the experiment in the dataset.
- Low alexithymia
- Group: 0
- TAS score: < 51
- Medium alexithymia
- Group: 1
- TAS score: 51 <= 60
- High alexithymia
- Group: 2
- TAS score: 61 <
id: Participant IDgender: Gender of participant; male (1), female (2)age: Age of participantTASDIF: TAS sub-score for difficulty identifying feelingsTASDDF: TAS sub-score for difficulty describing feelingsTASEOT: TAS sub-score for external-oriented thinkingTAStot: TAS total scoregroup: Label of alexithymia; low (0), medium (1), high (2)dataset: Label of dataset; dataset1 (1) and dataset2 (2)
| Dataset | Group Label | Gender | Mean Age | Mean TAS-score Total | Gender Count |
|---|---|---|---|---|---|
| 1 | L | M | 25 | 36 | 4 |
| 1 | L | F | 21 | 42 | 21 |
| 1 | H | M | 23 | 65 | 4 |
| 1 | H | F | 21 | 66 | 21 |
| 2 | L | M | 22 | 40 | 12 |
| 2 | L | F | 21 | 38 | 32 |
| 2 | M | M | 21 | 54 | 1 |
| 2 | M | F | 21 | 55 | 8 |
| 2 | H | M | 21 | 63 | 3 |
| 2 | H | F | 20 | 65 | 8 |
Group Label - L: Low, M: Medium, H: High
Gender - M: Male, F: Female
12113, 12116, 12125, 32115, 12010: Assigns specific values to the "Emotion_Tone_Phase" column based on a predetermined order of experiments, ensuring that the data is correctly labeled according to the experimental conditions.12236, 12001, 12004, 32138, 122306, 122347: Filter out unnecessary rows from the "Emotion_Tone_Phase" column based on a predetermined order of experiments, ensuring that the data is correctly labeled according to the experimental conditions.
- Fork the GitHub repository containing the Alexithymia Database Instructions to your own GitHub account. This will create a copy of the repository under your account.
- Clone the forked repository to your local machine using Git. Open a terminal or command prompt and run the following command:
git clone https://github.com/CaSToRC-CyI/ACII2023.git- Navigate to the cloned repository on your local machine, for example:
cd /Users/user/Documents/GitHub/ACII2023- Ensure that you have Python installed on your system. You can download Python from the official website be clicking here
- Install the necessary libraries by running the following command in your command prompt or terminal:
pip install bioread
pip install pandas
pip install numpy
pip install scipy
pip install -U scikit-learn
pip install tsfresh- Download the datasets upon request and save them in the directory in a folder called ACII2023_Datasets. Contact X for further instructions.
Structure of the ACII2023_Datasets folder:
- Dataset 1:
- LowAlexithymics
- HighAlexithymics
- Dataset 2:
- LowAlexithymics
- MediumAlexithymics
- HighAlexithymics
- Load the files and/or lists using the scripts found in the following links:
- Dataset 1:
- Dataset 2:
Read the csv file which contains the group label and the dataset label.
# Import libraries
from acii2023_functions import create_info_dictionary
# CSV file path
csv_file = '/Users/user/Desktop/ACII2023/ACII2023_Datasets/alexithymia_demographics_info.csv'
# Dictionary containing information for each participant
info_dict = create_info_dictionary(csv_file, 'id', ['group', 'dataset', 'TASDIF', 'TASDDF', 'TASEOT', 'TAStot'])Replace csv_file with your actual csv path.
Load data of individual participant from dataset 1 with high alexithymia:.
# Import libraries
from load_data import load_participant_data
# Folder path
folder_path = '/Users/user/Desktop/ACII2023/ACII2023_Datasets/Dataset1/HighAlexithymics/'
# Dictionary of participant 11251
participant_11251 = load_participant_data(folder_path, '11251.acq', info_dict, normalise=True, downsample=True, state='Universal')Replace folder_path with your actual path.
Note: The phase2 channel has been sampled at 125 Hz. In the above code, phase2 has been rescaled to be 1000 Hz as the other columns.
Load data of the participants of interest and start your analysis
Returns a dictionary of dictionaries which include physiological signals and information associated with specific participant
# Import libraries
from load_data import load_participant_data
# Folder path
folder_path = '/Users/user/Desktop/ACII2023/ACII2023_Datasets/Dataset1/HighAlexithymics/'
# List of selected participants to load dataset
dataset1_high = ['32115.acq', '22258.acq', '11254.acq', '12239.ACQ', '22249.acq', '32259.acq', '11251.acq']
X_1_high, y_1_high = load_dataset(folder_path, dataset1_high, info_dict, normalise=True, downsample=True, state='Universal', tsfresh=False)Returns dataframe with concatenated signals and dictionary of group labels
# Import libraries
from load_data import load_participant_data
# Folder path
folder_path = '/Users/user/Desktop/ACII2023/ACII2023_Datasets/Dataset1/HighAlexithymics/'
# List of selected participants to load dataset
dataset1_high = ['32115.acq', '22258.acq', '11254.acq', '12239.ACQ', '22249.acq', '32259.acq', '11251.acq']
X_1_high, y_1_high = load_dataset(folder_path, dataset1_high, info_dict, normalise=True, downsample=True, state='Universal', tsfresh=True)Check the sample script for entire analysis, including loading dataset, extracting tsfresh features, and performing logistic regression.
The database was collected from the Psychology department of the University of Cyprus.
The database will be publicly available upon request from the Cyprus Institute. You can fill the following form to request the data by clicking here.
If you use this database in your research or analysis, kindly cite it as follows:
Dataset 1:
Valeria Filippou, Nikolas Theodosiou, Mihalis Nicolaou, Elena Constantinou, Georgia Panayiotou, and Marios Theodorou. 2022. A Wavelet-based Approach for Multimodal Prediction of Alexithymia from Physiological Signals. In Companion Publication of the 2022 International Conference on Multimodal Interaction (ICMI '22 Companion). Association for Computing Machinery, New York, NY, USA, 177–184. https://doi.org/10.1145/3536220.3558076
Dataset 1 and 2:
Valeria Filippou, Mihalis A Nicolaou, Nikolas Theodosiou, Georgia Panayiotou, Elena Constantinou, Marios Theodorou and Panteli Maria. 2023. Multimodal Prediction of Alexithymia from Physiological and Audio Signals. 2023 11th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW)
This dataset is provided under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. By using this dataset, you agree to the terms and conditions specified in the license.
For any questions, concerns, or inquiries regarding the Alexithymia Database, please contact [v.filippou22@gmail.com].