Feature Extraction and Selection for Pain Recognition Using Peripheral Physiological Signals

Abstract

In pattern recognition, the selection of appropriate features is paramount to both the performance and the robustness of the system. Over-reliance on machine learning-based feature selection methods can, therefore, be problematic; especially when conducted using small snapshots of data. The results of these methods, if adopted without proper interpretation, can lead to sub-optimal system design or worse, the abandonment of otherwise viable and important features. In this work, a deep exploration of pain-based emotion classification was conducted to better understand differences in the results of the related literature. In total, 155 different time domain and frequency domain features were explored, derived from electromyogram (EMG), skin conductance levels (SCL), and electrocardiogram (ECG) readings taken from the 85 subjects in response to heat-induced pain. To address the inconsistency in the optimal feature sets found in related works, an exhaustive and interpretable feature selection protocol was followed to obtain a generalizable feature set. Associations between features were then visualized using a topologically-informed chart, called Mapper, of this physiological feature space, including synthesis and comparison of results from previous literature. This topological feature chart was able to identify key sources of information that led to the formation of five main functional feature groups: signal amplitude and power, frequency information, nonlinear complexity, unique, and connecting. These functional groupings were used to extract further insight into observable autonomic responses to pain through a complementary statistical interaction analysis. From this chart, it was observed that EMG and SCL derived features could functionally replace those obtained from ECG. These insights motivate future work on novel sensing modalities, feature design, deep learning approaches, and dimensionality reduction techniques.

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

“…In [278], chronic pain using heart rate and blood pressure sensors was evaluated. Signals of skin conductance levels, EMG and ECG were fused in [51] to analyze the response to thermal pain induced in healthy patients.…”

“…A preprocessing is applied to physiological signals to remove unwanted artifacts [51]. This preprocessing can be based on band-pass filters to remove motion artifacts and drifts from the continuous level of the signal, associated with the low-frequency components [49,50], or noise, normally included in the high-frequency components [35,50].…”

“…In the feature extraction stage, an abstraction of the relevant information from larger data elements of the physiological signal is performed [51,251,275]. The objective of this stage is to improve the density of information.…”

“…It should be noted that despite the lack of an objective reference for the evaluation of chronic pain, many studies have tried to induce pain levels in a controlled manner (by heat [23,[41][42][43][44][45][46][47][48][49][50][51][52][53], cold [45,48,52], pressure [44] or electrical stimulation [54]) as a first approximation to a method that allows the evaluation of chronic pain [1]. In contrast to the studies based on research with phasic pain stimuli (short duration), the number of studies based on tonic pain stimuli is growing as an approach to chronic pain [23,[41][42][43][44][45][46][47][48][49][50][51][52][52][53][54][55]. This validation approach represents an objective way for assessing chronic pain, leaving aside all the subjectivity that may be related to the perception of pain by the chronic patient [23,[41][42][43][44][45][46]…”

“…In contrast to the studies based on research with phasic pain stimuli (short duration), the number of studies based on tonic pain stimuli is growing as an approach to chronic pain [23,[41][42][43][44][45][46][47][48][49][50][51][52][52][53][54][55]. This validation approach represents an objective way for assessing chronic pain, leaving aside all the subjectivity that may be related to the perception of pain by the chronic patient [23,[41][42][43][44][45][46][47][48][49][50][51][52][52][53][54][55]. In addition, it becomes the most appropriate option in the case of patients with absence of communication skills, whether verbal, written or tactile (infants, unconscious patients, critically sick, patients with intellectual impairment, advanced dementia, etc.…”