Classification of Game Demand and the Presence of Experimental Pain Using Functional Near-Infrared Spectroscopy

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Pain Research

Background:

• Active distractions, like computer games, can enhance pain tolerance.
• This pain modulation is influenced by the cognitive demand of the distraction, with higher demand correlating to increased tolerance.

Purpose of the Study:

• To classify game demand levels and pain presence using neuroimaging (fNIRS) and physiological (ECG) data.
• To investigate the efficacy of machine learning models in differentiating between low and high game demand and the presence or absence of pain.

Main Methods:

• Twenty participants engaged in a racing game under varying demand (Easy/Hard) and pain (cold pressor test) conditions.
• Functional Near-InfraRed Spectroscopy (fNIRS) and electrocardiogram (ECG) data were collected.
• Machine learning classifiers (SVM, kNN, NB, RF) were trained and validated using selected fNIRS and ECG features.

Main Results:

• The Support Vector Machine (SVM) model, utilizing fNIRS data, accurately classified game demand levels (accuracy = 0.66, F1 = 0.68).
• Classifying the presence or absence of pain using the selected features and methods was not possible above chance levels.
• fNIRS data proved viable for classifying game demand, but pain classification remained challenging.

Conclusions:

• fNIRS is a promising tool for assessing cognitive load and game demand during interactive tasks.
• Detecting experimental pain is difficult when participants are engaged in a cognitively demanding task.
• Future research should explore advanced signal processing and multimodal approaches for pain detection during distraction.