Chronic pain classification using PPG and ECG parameters selected via hybrid feature selection

Area of Science:

• Biomedical Engineering
• Pain Medicine
• Signal Processing

Background:

• Current chronic pain assessment relies heavily on self-reporting, posing challenges for non-communicative patients.
• Medical staff often bear the burden of assessing pain in patients lacking communication ability or consciousness.
• There is a need for objective, automated methods to assess pain intensity in clinical settings.

Purpose of the Study:

• To develop an automated pain intensity classification system using physiological signals.
• To evaluate the effectiveness of photoplethysmography (PPG) and heart rate variability (HRV) features for pain assessment.
• To establish a reliable method for pain intensity classification in patients unable to self-report.

Main Methods:

• Collected electrocardiogram (ECG) and PPG recordings from 43 chronic pain patients and 20 healthy volunteers.
• Utilized a hybrid feature selection (HFS) method to identify key PPG and HRV indicators.
• Employed a multi-class support vector machine (SVM) with 10-fold cross-validation for pain classification into four levels: no, low, moderate, and severe pain.

Main Results:

• The SVM model achieved an overall estimation accuracy of approximately 74.6% for the four pain intensity levels.
• Hybrid feature selection identified robust PPG and HRV features capable of discriminating pain intensities.
• Analysis revealed novel trends in PPG features relevant to clinical pain assessment.

Conclusions:

• PPG and HRV features, combined with HFS, provide sufficient information to differentiate pain intensities in myofascial chronic pain patients.
• The developed system offers a promising automated approach for pain assessment in non-communicative individuals.
• This research contributes a new perspective on utilizing PPG signal trends for clinical pain evaluation.