Objectively measuring pain using facial expression: is the technology finally ready?

Artificial intelligence model predicting postoperative pain using facial expressions: a pilot study

PURPOSE

This study aimed to assess whether an artificial intelligence model based on facial expressions can accurately predict significant postoperative pain.

METHODS

A total of 155 facial expressions from patients who underwent gastric cancer surgery were analyzed to extract facial action units (AUs), gaze, landmarks, and positions. These features were used to construct various machine learning (ML) models, designed to predict significant postoperative pain intensity (NRS ≥ 7) from less significant pain (NRS < 7). Significant AUs predictive of NRS ≥ 7 were determined and compared to AUs known to be associated with pain in awake patients. The area under the receiver operating characteristic curves (AUROCs) of the ML models was calculated and compared using DeLong's test.

RESULTS

AU17 (chin raising) and AU20 (lip stretching) were found to be associated with NRS ≥ 7 (both P ≤ 0.004). AUs known to be associated with pain in awake patients did not show an association with pain in postoperative patients. An ML model based on AU17 and AU20 demonstrated an AUROC of 0.62 for NRS ≥ 7, which was inferior to a model based on all AUs (AUROC = 0.81, P = 0.006). Among facial features, head position and facial landmarks proved to be better predictors of NRS ≥ 7 (AUROC, 0.85-0.96) than AUs. A merged ML model that utilized gaze and eye landmarks, as well as head position and facial landmarks, exhibited the best performance (AUROC, 0.90) in predicting significant postoperative pain.

CONCLUSION

ML models using facial expressions can accurately predict the presence of significant postoperative pain and have the potential to screen patients in need of rescue analgesia.

TRIAL REGISTRATION NUMBER

This study was registered at ClinicalTrials.gov (NCT05477303; date: June 17, 2022).

Postoperative accurate pain assessment of children and artificial intelligence: A medical hypothesis and planned study

Although the pediatric perioperative pain management has been improved in recent years, the valid and reliable pain assessment tool in perioperative period of children remains a challenging task. Pediatric perioperative pain management is intractable not only because children cannot express their emotions accurately and objectively due to their inability to describe physiological characteristics of feeling which are different from those of adults, but also because there is a lack of effective and specific assessment tool for children. In addition, exposure to repeated painful stimuli early in life is known to have short and long-term adverse sequelae. The short-term sequelae can induce a series of neurological, endocrine, cardiovascular system stress related to psychological trauma, while long-term sequelae may alter brain maturation process, which can lead to impair neurodevelopmental, behavioral, and cognitive function. Children's facial expressions largely reflect the degree of pain, which has led to the developing of a number of pain scoring tools that will help improve the quality of pain management in children if they are continually studied in depth. The artificial intelligence (AI) technology represented by machine learning has reached an unprecedented level in image processing of deep facial models through deep convolutional neural networks, which can effectively identify and systematically analyze various subtle features of children's facial expressions. Based on the construction of a large database of images of facial expressions in children with perioperative pain, this study proposes to develop and apply automatic facial pain expression recognition software using AI technology. The study aims to improve the postoperative pain management for pediatric population and the short-term and long-term quality of life for pediatric patients after operational event.

Corrugator Muscle Activity Associated with Pressure Pain in Adults with Neck/Shoulder Pain

Background and Objectives: No studies have reported corrugator muscle activity associated with pain in people with pain. This study aimed to develop an objective pain assessment method using corrugator muscle activity with pressure pain stimulation to the skeletal muscle. Methods: Participants were 20 adults (a mean ± SD age of 22.0 ± 3.1 years) with chronic neck/shoulder pain. Surface electromyography (sEMG) of corrugator muscle activity at rest (baseline) and without and with pressure pain stimulation applied to the most painful tender point in the shoulder was recorded. Participants evaluated the intensity of the neck/shoulder pain and the sensory and affective components of pain with pressure stimulation using a visual analogue scale (VAS). The percentages of integrated sEMG (% corrugator activity) without and with pressure pain stimulation to the baseline integrated sEMG were compared, and the relationships between the % corrugator activity and the sensory and affective components of pain VAS scores were evaluated. Results: Without pressure stimulation, an increase in corrugator muscle activity due to chronic neck/shoulder pain was not observed. The % corrugator activity with pressure pain stimulation was significantly higher than that without stimulation (p < 0.01). A significant positive correlation between corrugator muscle activity and the affective components of pain VAS scores with pressure stimulation was found (ρ = 0.465, p = 0.039) and a tendency of positive correlation was found for the sensory component of pain VAS scores (ρ = 0.423, p = 0.063). Conclusions: The increase in corrugator muscle activity with pressure pain stimulation to the tender point in adults with chronic neck/shoulder pain was observed, although increased corrugator muscle activity resulting from the chronic neck/shoulder pain was not. These findings suggest that corrugator muscle activity with pressure pain stimulation can be a useful objective indication for tender point sensitivity assessment in the skeletal muscle with pain.

Perspectives on pain in Down syndrome

Down syndrome (DS) or trisomy 21 is a genetic condition often accompanied by chronic pain caused by congenital abnormalities and/or conditions, such as osteoarthritis, recurrent infections, and leukemia. Although DS patients are more susceptible to chronic pain as compared to the general population, the pain experience in these individuals may vary, attributed to the heterogenous structural and functional differences in the central nervous system, which might result in abnormal pain sensory information transduction, transmission, modulation, and perception. We tried to elaborate on some key questions and possible explanations in this review. Further clarification of the mechanisms underlying such abnormal conditions induced by the structural and functional differences is needed to help pain management in DS patients.

Artificial Intelligence for Automatic Pain Assessment: Research Methods and Perspectives

Although proper pain evaluation is mandatory for establishing the appropriate therapy, self-reported pain level assessment has several limitations. Data-driven artificial intelligence (AI) methods can be employed for research on automatic pain assessment (APA). The goal is the development of objective, standardized, and generalizable instruments useful for pain assessment in different clinical contexts. The purpose of this article is to discuss the state of the art of research and perspectives on APA applications in both research and clinical scenarios. Principles of AI functioning will be addressed. For narrative purposes, AI-based methods are grouped into behavioral-based approaches and neurophysiology-based pain detection methods. Since pain is generally accompanied by spontaneous facial behaviors, several approaches for APA are based on image classification and feature extraction. Language features through natural language strategies, body postures, and respiratory-derived elements are other investigated behavioral-based approaches. Neurophysiology-based pain detection is obtained through electroencephalography, electromyography, electrodermal activity, and other biosignals. Recent approaches involve multimode strategies by combining behaviors with neurophysiological findings. Concerning methods, early studies were conducted by machine learning algorithms such as support vector machine, decision tree, and random forest classifiers. More recently, artificial neural networks such as convolutional and recurrent neural network algorithms are implemented, even in combination. Collaboration programs involving clinicians and computer scientists must be aimed at structuring and processing robust datasets that can be used in various settings, from acute to different chronic pain conditions. Finally, it is crucial to apply the concepts of explainability and ethics when examining AI applications for pain research and management.

The neurobiology of pain and facial movements in rodents: Clinical applications and current research

One of the most controversial aspects of the use of animals in science is the production of pain. Pain is a central ethical concern. The activation of neural pathways involved in the pain response has physiological, endocrine, and behavioral consequences, that can affect both the health and welfare of the animals, as well as the validity of research. The strategy to prevent these consequences requires understanding of the nociception process, pain itself, and how assessment can be performed using validated, non-invasive methods. The study of facial expressions related to pain has undergone considerable study with the finding that certain movements of the facial muscles (called facial action units) are associated with the presence and intensity of pain. This review, focused on rodents, discusses the neurobiology of facial expressions, clinical applications, and current research designed to better understand pain and the nociceptive pathway as a strategy for implementing refinement in biomedical research.

ASAD: A Novel Audification Console for Assessment and Communication of Pain and Discomfort

Pain and discomfort are subjective perceptions that are difficult to quantify. Various methods and scales have been developed to find an optimal manner to describe them; however, these are difficult to use with some categories of patients. Audification of pain has been utilized as feedback in rehabilitation settings to enhance motor perception and motor control, but not in assessment and communication settings. We present a novel tool, the Audification-console for Self-Assessment of Discomfort (ASAD), for assessing and communicating pain and discomfort through sound. The console is a sequence of increasing pitch and frequencies triggered at the press of buttons and displayed as a matrix that can be associated with the subjective perception of pain and discomfort. The ASAD has been evaluated in its ability to capture and communicate discomfort, following a fatigue test in the lower limbs with thirty healthy volunteers, and compared to the most common self-reported methods used in the NHS. (The National Health Service (NHS) is the publicly funded healthcare system in England and one of the four National Health Service systems in the United Kingdom.) This was a qualitative, within subjects and across groups experiment study. The console provides a more accurate assessment than other scales and clearly recognizable patterns of sounds, indicating increased discomfort, significantly localized in specific frequency ranges, thus easily recognizable across subjects and in different instances of the same subject. The results suggest a possible use of the ASAD for a more precise and automatic assessment of pain and discomfort in health settings. Future studies might assess if this is easier to use for patients with communication or interpretation difficulties with the traditional tools.

Novel Method for Three-Dimensional Facial Expression Recognition Using Self-Normalizing Neural Networks and Mobile Devices

Introduction To date, most ways to perform facial expression recognition rely on two-dimensional images, advanced approaches with three-dimensional data exist. These however demand stationary apparatuses and thus lack portability and possibilities to scale deployment. As human emotions, intent and even diseases may condense in distinct facial expressions or changes therein, the need for a portable yet capable solution is signified. Due to the superior informative value of three-dimensional data on facial morphology and because certain syndromes find expression in specific facial dysmorphisms, a solution should allow portable acquisition of true three-dimensional facial scans in real time. In this study we present a novel solution for the three-dimensional acquisition of facial geometry data and the recognition of facial expressions from it. The new technology presented here only requires the use of a smartphone or tablet with an integrated TrueDepth camera and enables real-time acquisition of the geometry and its categorization into distinct facial expressions. Material and Methods Our approach consisted of two parts: First, training data was acquired by asking a collective of 226 medical students to adopt defined facial expressions while their current facial morphology was captured by our specially developed app running on iPads, placed in front of the students. In total, the list of the facial expressions to be shown by the participants consisted of "disappointed", "stressed", "happy", "sad" and "surprised". Second, the data were used to train a self-normalizing neural network. A set of all factors describing the current facial expression at a time is referred to as "snapshot". Results In total, over half a million snapshots were recorded in the study. Ultimately, the network achieved an overall accuracy of 80.54% after 400 epochs of training. In test, an overall accuracy of 81.15% was determined. Recall values differed by the category of a snapshot and ranged from 74.79% for "stressed" to 87.61% for "happy". Precision showed similar results, whereas "sad" achieved the lowest value at 77.48% and "surprised" the highest at 86.87%. Conclusions With the present work it can be demonstrated that respectable results can be achieved even when using data sets with some challenges. Through various measures, already incorporated into an optimized version of our app, it is to be expected that the training results can be significantly improved and made more precise in the future. Currently a follow-up study with the new version of our app that encompasses the suggested alterations and adaptions, is being conducted. We aim to build a large and open database of facial scans not only for facial expression recognition but to perform disease recognition and to monitor diseases' treatment progresses.

Self-Reported Efficacy of Treatments in Cluster Headache: a Systematic Review of Survey Studies

PURPOSE OF REVIEW

The use and efficacy of various substances in the treatment of CH have been studied in several retrospective surveys. The aim of the study is to systematically review published survey studies to evaluate the reported efficacies of both established and unconventional substances in abortive and prophylactic treatment of both episodic and chronic CH, specifically assessing the consistency of the results.

RECENT FINDINGS

No systematic review have been conducted of these studies previously. A systematic literature search with a set of search terms was conducted on PubMed. Retrospective surveys that quantified the self-reported efficacy of two or more CH treatments, published in English during 2000-2020, were included. Several key characteristics and results of the studies were extracted. A total of 994 articles were identified of which 9 were found to be eligible based on the selection criteria. In total, 5419 respondents were included. Oxygen and subcutaneous triptan injections were most reported as effective abortive treatments, while psilocybin and lysergic acid diethylamide were most commonly reported as effective prophylactic treatments. The reported efficacy of most substances was consistent across different studies, and there were marked differences in the reported efficacies of different substances. The reported order of efficacy is generally in agreement with clinical studies. The findings suggest that retrospective surveys can be used to obtain supporting information on the effects of various substances used in the treatment of CH and to form hypotheses about novel treatment methods. The consistently reported efficacy of psilocybin and LSD in prophylactic treatment indicates need for clinical studies.

The Current View on the Paradox of Pain in Autism Spectrum Disorders

Autism spectrum disorder (ASD) is a neurodevelopmental disorder, which affects 1 in 44 children and may cause severe disabilities. Besides socio-communicational difficulties and repetitive behaviors, ASD also presents as atypical sensorimotor function and pain reactivity. While chronic pain is a frequent co-morbidity in autism, pain management in this population is often insufficient because of difficulties in pain evaluation, worsening their prognosis and perhaps driving higher mortality rates. Previous observations have tended to oversimplify the experience of pain in autism as being insensitive to painful stimuli. Various findings in the past 15 years have challenged and complicated this dogma. However, a relatively small number of studies investigates the physiological correlates of pain reactivity in ASD. We explore the possibility that atypical pain perception in people with ASD is mediated by alterations in pain perception, transmission, expression and modulation, and through interactions between these processes. These complex interactions may account for the great variability and sometimes contradictory findings from the studies. A growing body of evidence is challenging the idea of alterations in pain processing in ASD due to a single factor, and calls for an integrative view. We propose a model of the pain cycle that includes the interplay between the molecular and neurophysiological pathways of pain processing and it conscious appraisal that may interfere with pain reactivity and coping in autism. The role of social factors in pain-induced response is also discussed. Pain assessment in clinical care is mostly based on subjective rather than objective measures. This review clarifies the strong need for a consistent methodology, and describes innovative tools to cope with the heterogeneity of pain expression in ASD, enabling individualized assessment. Multiple measures, including self-reporting, informant reporting, clinician-assessed, and purely physiological metrics may provide more consistent results. An integrative view on the regulation of the pain cycle offers a more robust framework to characterize the experience of pain in autism.

Real-Time Classification of Pain Level Using Zygomaticus and Corrugator EMG Features

The real-time recognition of pain level is required to perform an accurate pain assessment of patients in the intensive care unit, infants, and other subjects who may not be able to communicate verbally or even express the sensation of pain. Facial expression is a key pain-related behavior that may unlock the answer to an objective pain measurement tool. In this work, a machine learning-based pain level classification system using data collected from facial electromyograms (EMG) is presented. The dataset was acquired from part of the BioVid Heat Pain database to evaluate facial expression from an EMG corrugator and EMG zygomaticus and an EMG signal processing and data analysis flow is adapted for continuous pain estimation. The extracted pain-associated facial electromyography (fEMG) features classification is performed by K-nearest neighbor (KNN) by choosing the value of k which depends on the nonlinear models. The presentation of the accuracy estimation is performed, and considerable growth in classification accuracy is noticed when the subject matter from the features is omitted from the analysis. The ML algorithm for the classification of the amount of pain experienced by patients could deliver valuable evidence for health care providers and aid treatment assessment. The proposed classification algorithm has achieved a 99.4% accuracy for classifying the pain tolerance level from the baseline (P0 versus P4) without the influence of a subject bias. Moreover, the result on the classification accuracy clearly shows the relevance of the proposed approach.

Facial Muscles Reactions to Other Person’s Facial Expressions of Pain

The aim of this study was to record facial electromiograms (EMG) while subjects were viewing facial expressions of different pain levels (no-pain, medium pain and very painful) and to find objective criteria for measuring pain expressed in human’s face. The study involved 18 students with age 21 years. The magnitude of the EMG response of m. corrugator supercilii depended on voluntary performed facial pain expression in the subjects. EMG responses of voluntary performed facial pain expressions to mirrored pain reactions were detected at two time span intervals: 200–300 ms after stimulation in m. zygomaticus major, and 400–500 ms after stimulation in m. corrugator supercilii. These differences disappear after 1300 ms. In the second time interval, differences in EMG responses of both muscle groups occur 1600 ms after stimulus presentation, but disappear differently: 3100 ms after stimulation in m. zygomaticus major and 4000 ms in m. corrugator supercilii. Constant responding with “medium pain” expression when recognizing faces of different pain expressions have an effect on the voluntary EMG responses of individual subjects. Images with emotional expression “no pain” reduce m. corrugator supercilii activity and increase m. zygomaticus major activity for those observers.

Evaluation of facial electromyographic pain responses in healthy participants

Aim: Assessing pain perception through self-reports may not be possible in some patients, for example, sedated. Our group considered if facial electromyography (fEMG) could provide a useful alternative, by testing on healthy participants subjected to experimental pain. Materials & methods: Activity of four facial muscles was recorded using fEMG alongside self-reported pain scores and physiological parameters. Results: The pain stimulus elicited significant activity on all facial muscles of interest as well as increases in heart rate. Activity from two of the facial muscles correlated significantly against pain intensity. Conclusion: Pain perception can be assessed through fEMG on healthy participants. We believe that this model would be valuable to clinicians that need to diagnose pain perception in circumstances where verbal reporting is not possible.

Developing a pain intensity prediction model using facial expression: A feasibility study with electromyography

The automatic detection of facial expressions of pain is needed to ensure accurate pain assessment of patients who are unable to self-report pain. To overcome the challenges of automatic systems for determining pain levels based on facial expressions in clinical patient monitoring, a surface electromyography method was tested for feasibility in healthy volunteers. In the current study, two types of experimental gradually increasing pain stimuli were induced in thirty-one healthy volunteers who attended the study. We used a surface electromyography method to measure the activity of five facial muscles to detect facial expressions during pain induction. Statistical tests were used to analyze the continuous electromyography data, and a supervised machine learning was applied for pain intensity prediction model. Muscle activation of corrugator supercilii was most strongly associated with self-reported pain, and the levator labii superioris and orbicularis oculi showed a statistically significant increase in muscle activation when the pain stimulus reached subjects' self -reported pain thresholds. The two strongest features associated with pain, the waveform length of the corrugator supercilii and levator labii superioris, were selected for a prediction model. The performance of the pain prediction model resulted in a c-index of 0.64. In the study results, the most detectable difference in muscle activity during the pain experience was connected to eyebrow lowering, nose wrinkling and upper lip raising. As the performance of the prediction model remains modest, yet with a statistically significant ordinal classification, we suggest testing with a larger sample size to further explore the variables that affect variation in expressiveness and subjective pain experience.

Artificial Intelligence and Machine Learning in Anesthesiology

Commercial applications of artificial intelligence and machine learning have made remarkable progress recently, particularly in areas such as image recognition, natural speech processing, language translation, textual analysis, and self-learning. Progress had historically languished in these areas, such that these skills had come to seem ineffably bound to intelligence. However, these commercial advances have performed best at single-task applications in which imperfect outputs and occasional frank errors can be tolerated.The practice of anesthesiology is different. It embodies a requirement for high reliability, and a pressured cycle of interpretation, physical action, and response rather than any single cognitive act. This review covers the basics of what is meant by artificial intelligence and machine learning for the practicing anesthesiologist, describing how decision-making behaviors can emerge from simple equations. Relevant clinical questions are introduced to illustrate how machine learning might help solve them-perhaps bringing anesthesiology into an era of machine-assisted discovery.

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

Non-oncologic chronic pain is a common high-morbidity impairment worldwide and acknowledged as a condition with significant incidence on quality of life. Pain intensity is largely perceived as a subjective experience, what makes challenging its objective measurement. However, the physiological traces of pain make possible its correlation with vital signs, such as heart rate variability, skin conductance, electromyogram, etc., or health performance metrics derived from daily activity monitoring or facial expressions, which can be acquired with diverse sensor technologies and multisensory approaches. As the assessment and management of pain are essential issues for a wide range of clinical disorders and treatments, this paper reviews different sensor-based approaches applied to the objective evaluation of non-oncological chronic pain. The space of available technologies and resources aimed at pain assessment represent a diversified set of alternatives that can be exploited to address the multidimensional nature of pain.

Pain in dementia

The ageing revolution is changing the composition of our society with more people becoming very old with higher risks for developing both pain and dementia. Pain is normally signaled by verbal communication, which becomes more and more deteriorated in people with dementia. Thus, these individuals unnecessarily suffer from manageable but unrecognized pain. Pain assessment in patients with dementia is a challenging endeavor, with scientific advancements quickly developing. Pain assessment tools and protocols (mainly observational scales) have been incorporated into national and international guidelines of pain assessment in aged individuals. To effectively assess pain, interdisciplinary collaboration (nurses, physicians, psychologists, computer scientists, and engineers) is essential. Pain management in this vulnerable population is also preferably done in an interdisciplinary setting. Nonpharmacological management programs have been predominantly tested in younger populations without dementia. However, many of them are relatively safe, have proven their efficacy, and therefore deserve a first place in pain management programs. Paracetamol is a relatively safe and effective first-choice analgesic. There are many safety issues regarding nonsteroidal anti-inflammatory drugs, opioids, and adjuvant analgesics in dementia patients. It is therefore recommended to monitor both pain and potential side effects regularly. More research is necessary to provide better guidance for pain management in dementia.

The utility of adding symptoms and signs to the management of injury-related pain

OBJECTIVE

Improved pain assessment and management in the emergency department (ED) is warranted. We aimed to determine the impact on pain management, of adding symptoms and signs to pain assessment.

PATIENTS AND METHODS

A single center before-and-after study was conducted, supplemented by an interrupted time series analysis. The intervention included the addition of clinical presentation (CP) of the injury and facial expression (FE) of the patient to pain assessment scales of patients with soft tissue injures. Pain intensity was categorized as: mild, moderate, and severe. We compared types of pain relief medications, use of strong opioids, and pain relief efficacy between pre and post intervention phases.

RESULTS

Before-and-after analysis revealed a significant reduction in the use of strong opioids. The adjusted relative ratio for the use of strong opioids in the post intervention phase was 0.63 (95% CI: 0.48-0.82). This reduction was mostly driven by less use of strong opioids in patients reporting severe pain (from 17.3%-7.9%) (P < 0.0001). A larger proportion of patients in the post intervention phase than in the pre intervention phase received weak opioids and nonsteroidal anti-inflammatory drugs (NSAIDs) (27.4% vs 19.1%, P = 0.002), and a larger proportion did not receive any pain relief medication (19.8% vs 10.5%, p < 0.0001). The use of strong opioids increased with higher levels of FE and CP. Among patients with mild injury and reporting severe pain, the odds of receiving a strong opioid was nearly 9 times (OR = 8.9, 95% CI: 4.0-19.6) higher among those who were with an unrelaxed FE and showed pain behavior than those with relaxed FE. Interrupted time-series analysis showed that the mean ΔVAS (VAS score at entry minus VAS score at discharge) in the post intervention phase compared with the pre intervention phase was not statistically significant (P = 0.073). The use of strong opioids in the post intervention phase was significantly reduced (P = 0.017).

CONCLUSION

Adding symptoms and signs to pain assessment of patients admitted with soft tissue injuries decreased the use of strong opioids, without affecting pain relief efficacy.