Objective pain stimulation intensity and pain sensation assessment using machine learning classification and...

Area of Science:

• Biomedical Engineering
• Neuroscience
• Machine Learning Applications

Background:

• Chronic pain affects a significant portion of the adult population in the United States, necessitating objective pain measurement tools.
• Current pain quantification relies on subjective self-reporting (e.g., Visual Analog Scale - VAS), complicating management and contributing to the opioid crisis.
• Electrodermal activity (EDA) offers a potential physiological marker for objective pain assessment.

Purpose of the Study:

• To investigate the efficacy of machine learning models in objectively estimating pain sensation using electrodermal activity (EDA).
• To differentiate between pain stimulation intensity and subjective pain sensation using physiological data.
• To explore the correlation between EDA responses and applied stimuli versus self-reported pain levels.

Main Methods:

• Twenty-three healthy volunteers were subjected to controlled thermal grill pain stimulation at three intensity levels.
• Electrodermal activity (EDA) data were collected throughout the experiment, and validated features were extracted.
• Classification and regression machine learning models were employed to estimate pain stimulation intensity and self-reported pain sensation (VAS).

Main Results:

• Machine learning models achieved macroaveraged geometric mean scores of 69.7% for pain stimulation intensity and 69.2% for pain sensation (three-class classification).
• Regression models yielded R-squared values of 0.357 for stimulation intensity and 0.47 for pain sensation.
• EDA responses showed a stronger correlation with the intensity of applied stimuli than with the subjectively reported pain sensation, despite VAS score variability.

Conclusions:

• Objective quantification of three distinct pain levels is achievable with good accuracy using EDA and machine learning.
• EDA-based sympathetic responses are more closely aligned with the physical intensity of pain stimuli than with subjective pain perception.
• This research provides physiological evidence supporting EDA as a potential objective biomarker for pain assessment, complementing subjective measures.