Data Analysis Pipeline (HCI-SENSE-42-1.0.0-analysis.zip)

The dataset was collected from 42 participants over a 2-hour continuous interaction session, during which participants engaged in a series of designed tasks on a desktop computer with a mouse and keyboard. The experimental tasks were conducted within a fully simulated desktop operating system environment, designed to closely mimic real-world computer usage scenarios. This setup mirrors how people typically use computers in daily life. The simulated experiment program also enables the comprehensive capture of the mouse and keyboard data, synchronised with a high refresh rate monitor at 144 Hz. Recordings were collected in a noise-insulated room, minimising external interruptions from the environment or the experimenter.

The dataset includes multimodal data, including

• Behavioural Data
• 32-channel Electroencephalogram Recordings with BioSemi ActiveTwo System
• Respiratory Cycles with Respiration Belt
• 3-lead Electrocardiogram Recordings*
• Webcam Recordings*

[*] published data modality is limited to participants consented for sharing them to the public

Repository Structure

├── README.md  # This README file
├── LICENSE    # CC0 License file
├── data    # Location for downloaded dataset files
│   ├── 001_explorer_2025-02-15_15h23.13.921.csv
│   ├── 001_explorer_2025-02-15_15h23.13.921.psydat
│   ├── ...  # More behavioural data files expected here
│   ├── ECG  # ECG recording files (*.fif)
│   │   ├── P002.fif
│   │   ├── ...
│   ├── EEG  # EEG recording files (*.bdf)
│   │   ├── P001.bdf
│   │   ├── ...
│   ├── Respiration  # Respiratory files (*.wav)
│   │   ├── P001.wav
│   │   ├── ...
│   ├── Events.txt   # Definition of EEG events
│   └── participant_enrollment.csv  # Encoded participant information
├── analyze_*  # Contains Python scripts for data analysis
│   ├── check_*.py    # Basic parsing and visualization script
│   ├── convert_*.py  # Data parsing and format conversion script
│   ├── ...
├── condition  # Analyze the data with the condition given
│   ├── with_comfort
│   ├── with_handedness
│   ├── with_keyboard_frequency
│   ├── with_os
│   └── with_usage_hours
├── correlation  # Analyze the dataset by investigating the correlation
│   └── with_questionnaire
├── data_definition.py  # Definition of the list of *.psydat files
├── data_parser.py      # Definition of the *.psydat parsing class
├── main.py             # A small example script to parse for sleepiness levels across time domain
├── requirements.txt    # Python package requirement file
└── test_materials      # Analysis related to typing materials
    └── check_text_materials.py  # Parsing and visualization of the word distributions

💡 After downloaded the [HCI-SENSE-42 Dataset](https://doi.org/10.5281/zenodo.20328098), we expect them to be organized in the format listed above.

Getting Started

Prerequisites

• Python 3.9+ installed
• pip available
• RStudio installed (to run R scripts)

Clone this repository

git clone https://github.com/Catherine9811/HCI-SENSE-42.git
cd HCI-SENSE-42

Install the dependencies

pip install -r requirements.txt

💡 The `numpy` package version must match the `PsychoPy` package versions. In my case, numpy==1.26.4 and psychopy==2024.2.3

Run a quick test

python3 main.py

💡 For parsing and checking data for each modality, refer to scripts located in `analyze_*/check_*.py` to get started

Accessing the Dataset

The dataset is hosted on https://doi.org/10.5281/zenodo.20328098.

Data from sensors with different modalities are flattened to allow for separate downloads if not all of them are required in the analysis.

💡 You must login and agree to the conditions for use before accessing the webcam recordings as it contains sensitive information.