Automated Pain Assessment using Electrodermal Activity Data and Machine Learning

Abstract

Objective pain assessment is required for appropriate pain management in the clinical setting. However, clinical gold standard pain assessment is based on subjective methods. Automated pain detection from physiological data may provide important objective information to better standardize pain assessment. Specifically, electrodermal activity (EDA) can identify features of stress and anxiety induced by varying pain levels. However, notable variability in EDA measurement exists and research to date has demonstrated sensitivity but lack of specificity in pain assessment. In this paper, we use timescale decomposition (TSD) to extract salient features from EDA signals to identify an accurate and automated EDA pain detection algorithm to sensitively and specifically distinguish pain from no-pain conditions.

Specific Behavioral Responses Rather Than Autonomic Responses Can Indicate and Quantify Acute Pain among Individuals with Intellectual and Developmental Disabilities

“…Thus, although autonomic variables may be affected by acute clinical pain conditions and are easy to monitor; their use as pain indicators requires further investigation with stronger experimental stimuli than those used here, and perhaps with additional stimulation modalities. A new approach to analyze the electrodermal activity may hold promise pending further examination [ 43 ].…”

Sensor Technologies to Manage the Physiological Traits of Chronic Pain: A Review

“…• Support vector machine (SVM): In SVM model the data are represented as points in a hyper-space of high dimension. The optimization process tries to find the hyper-plane that defines the separation surface between categories, maximizing the distance from all points: [35,46,51,54,258,277,[283][284][285][286].…”

“…Table 1 shows a summary of some of the results obtained with machine learning techniques in pain assessment studies (pain classification accuracy defined according to [284]). [290] Musculoskeletal chronic pain RF Electronic health records 94% [244] Infant pain RVM Facial expressions (computer vision) 91% [252] Chronic low back pain SVM fMRI 92.5% [258] Fibromyalgia DT fMRI 76% [48] Induced pain in healthy people K-NN fNIRS 88.3% [216] Shoulder pain SVM EEG 84% [298] Multiple pain etiology CNN Facial expressions (computer vision) 93.3% [283] Pain after surgery SVM Facial expressions (computer vision) 87% [286] Pain after surgery SVM Skin conductance 77.7% [299] Shoulder pain CNN Facial expressions (computer vision) 98.5% [300] Infant pain CNN Facial expressions (computer vision) 88.3% [49] Induced pain in healthy people SVM EEG 78.2% [50] Induced pain in healthy people RF EEG 89.5% [277] Sickle cell disease pain k-NN, SVM Multisensor 68% [278] Multiple chronic pain RBM Heart rate, blood pressure 72% [52] Induced pain in healthy people CNN EEG 97.4% [51] Induced pain in healthy people SVM EMG, skin conductance, ECG 79.4% [54] Neck and shoulder pain SVM EMG 77%…”

Machine Learning for Stress Detection from Electrodermal Activity: A Scoping Review

“…It is a good practice to use the best features that are most suited in relation to their contrasting performance. In the time domain, the following features are commonly used: mean amplitude (Mean); amplitude standard deviation (SD), the SD first and second derivative (D1, D2), the SD means (D1M, D2M) and their standard deviations (D1SD and D2SD) [29]; sum rise time (SRT), sum fall time (SFT), rise rate mean (RM), rise rate standard deviation (RRSTD); decay rate mean (DCRM), decay rate standard deviation (DCRSD); phasic value mean (PHVM), phasic value standard deviation (PHVSD); startle time mean (STM), startle time standard deviation (STSD), startle RMS mean (STRMS), startle RMS standard deviation (STRMSSD); startle RMS overall (STRMSOV); electrodermal level (EDL), electrodermal response (EDR); cumulative maximum (CMax), cumulative minimum (CMin); smallest window elements (SWE); dynamic range (DR); root-mean square level (RMS), peak-magnitude-to-RMS ratio (PMRMSR); root-sum-of-squares level (RSSL); peak (P), peak location (PLoc), peak to peak time (PPT), peaks intervals differs 50ms (pNN50) [24,40,41,53,65,75].…”

“…A total of 20 papers have only used EDA signals for stress detection [17][18][19][20]24,[27][28][29][30][31][32][33]36,42,43,[45][46][47]49,53,73,103]. The authors have focused on developing methods or evaluating ML models based in EDA and its features.…”