To analyze the use of physiological signals and machine learning models in pain assessment across various pain types, including acute, chronic, and postoperative pain, and identify patterns in behavioral features.
Key Findings:
The most investigated physiological signals for pain assessment were electrocardiography (ECG), electrodermal activity (EDA), photoplethysmography (PPG), and electroencephalography (EEG). Notably, 36 studies employed machine learning approaches for pain assessment, while 53 focused on traditional statistical methods, highlighting a shift towards more advanced analytical techniques. Two commercial devices for pain monitoring were identified: the Medasense PMD-200 and the Newborn Infant Parasympathetic Evaluation (NIPE) Monitor, indicating practical applications of research findings.
Interpretation:
Despite the growing interest in objective pain assessment through physiological signals, the absence of a widely accepted method poses significant challenges for clinical practice, underscoring the need for further validation and standardization.
Limitations:
The review may not encompass all relevant studies due to the specific search strategy employed, which may have inadvertently excluded studies utilizing less common physiological signals.
Conclusion:
There is a need for further research to establish standardized methods for using physiological signals in pain assessment across various contexts, particularly focusing on integrating these methods into clinical practice.
