Analyzing Multimodal Signals with Wavelet Scattering for Pain Detection in Physiotherapy

Julian Grayson^*
*Correspondence: Julian Grayson, Department of Cardiothoracic Surgery, Presbyterian Hospital, New York, USA, Email:

Keywords

Physiotherapy • Pain monitoring • Pain assessment

Introduction

The sensory, emotional, cognitive, and social aspects of actual or potential tissue damage lead to the distressing sensation of pain. These aspects indicate that pain is subjective. The same stimulus with the same duration, source, and location can cause multiple reactions in patients. As a result, rather than using numerical measures like pain intensity ratings or pain threshold levels, pain should be evaluated in a personal context and with meaning. Self-reporting pain may also be difficult for some patients, such as young children, people with intellectual disabilities, critically ill or unconscious people, and adults with advanced dementia. Other sounds, gestures, head positions, and eye blinks can be used by patients to self-report pain in these situations, but the information they receive may be skewed by a clinician's lack of experience or uncertainty regarding the response. The difficulties of pain assessment led to the development of automatic pain monitoring systems.

Literature Review

There have been a number of studies on automatic pain monitoring systems. They differ based on the type of data collected, the cause of pain, and the method used to measure the patient's subjective pain rate. Numerous of them were based on freely available data sets. 90 healthy adults' data are included in the BioVid Heat Pain Database. The experiment was built on top of the calibration procedure and a carefully planned set of four-level thermal stimulation. The database contains the biopotentials listed below: electromyography of the corrugator, zygomaticus, and trapezius muscles (EMG), electrocardiography, and electrodermal activity (EDA) are all forms of electroencephalography. Additionally, depth maps and a video of the subjects' faces were obtained. An approach for continuously estimating pain intensity in a person-independent setting was presented in. The authors emphasized that the BioVid Databases' ability to recognize pain without requiring personalization is effective. Biosignals and video features were utilized. An algorithm for regression based on recurrent neural networks was presented in. A collection of various modalities was gathered for inclusion in the X-ITE Pain Database.

Along with ECG, EMG, EDA, and video of the face and body, thermal video data and audio signal are also included. Additionally, the survey utilized two sources of pain: heat and electrical stimuli. In, short/phasic stimulation was found to be more effective than long/tonic stimulation. The effectiveness of combining various modalities was also evaluated. Despite the relatively recent database (2019), the research is still ongoing. 129 people who had shoulder pain provided information to the UNBC-McMaster Shoulder Pain Database. The study plan assumed eight motion tests. During the passive tests, a physiotherapist moved the patient's limb, but during the active tests, the patient stood. The pain was evaluated using a self-report and an offline independent rating based on a scene video after the examination. Additionally, the face was captured on video. For pain intensity regression, the authors of this paper suggested using a CNN (convolutional neural network). The spatiotemporal information from face videos was used in. The authors compared the performance of descriptors on the BioVid and UNBC-McMaster databases [1,2].

EmoPain and UNBC-McMaster research appears to address the need for pain monitoring in physiotherapy. We were able to estimate the level of pain experienced by patients with shoulder pain during motion tests by utilizing face activity descriptors. For a two-class problem, the authors claim an accuracy of 92 percent. Both healthy participants and patients with chronic musculoskeletal pain were examined. The study protocol included exercises for sitting to standing and full trank flexion. The authors were able to identify three-class pain levels with an accuracy of approximately 70–95% by utilizing kinematic and electromyographic (EMG) features. It was discovered that people with chronic pain change how they exercise, such as by using different parts of their bodies for fear of hurting themselves. As a consequence of this, movements of the body reflected protective responses to pain. The emotional state (such as fear of pain, anxiety, or demotivation) strongly influences the efficacy of self-exercise-based therapy. There are numerous reasons why manual therapy is superior to physical exercise. The patient sets his or her own parameters for physical exercise, which may impact how bad the pain is.

Facial therapy relies heavily on the physiotherapist's intuition and experience because it is a direct intervention in the patient's tissue. Protective behaviors are not displayed by the patient because the patient has no control over the external cause of the pain. Additionally, it was observed that this particular procedure frequently results in unexpected pain. It makes fascial therapy inapplicable to the aforementioned methods for assessing pain. It is one of the treatments for musculoskeletal pain syndromes that are caused by overload (mostly because the position must be held for a long time). Its goal is to make the muscles more flexible and loosen up. These objectives are accomplished through slow movement along the muscle fibers and point pressure or deep rubbing. It improves blood flow by increasing tissue oxygenation. The physiotherapist transmits a customized set of variable stimuli through their hand in terms of strength and plane during the therapy course. During its performance, it is necessary to warm up tissues from the superficial to the deeper ones. This effect is produced by gradually increasing the strength and intensity of the stimulus. The range of force that could hurt the patient should not be exceeded during the procedure. In a therapy that is not supported by a pain monitoring system, the only measure that can provide real-time feedback on the intensity of the therapy is nociceptive pain information.

Discussion

Rapid behavioral responses like sighing, bracing, and grimacing are linked to this test. The risk of tissue damage and sensitive patient reactions in patients who are resistant to treatment could have a negative impact on the therapy's success. Consequently, using an automated pain assessment system may help ensure the safe and effective delivery of the therapy. Additionally, such a system may help keep the patient from becoming irritated by the constant requests regarding perceived pain intensity. As far as we are aware, no studies have been conducted on pain-related reaction monitoring during manual physiotherapy. The goal of this study is to create a pain-related reaction (PRR) assessment tool that can be used with facial therapy. During physiotherapy, a multimodal setup that was designed and implemented is used to acquire and synchronize patient data. Using an advanced signal analysis method and feature extraction in the time and frequency domain, a vector is subjected to a classification procedure that enables the detection of PRR level changes that may not be visible in every registered signal. This makes it possible to find such changes. In different patients, the painrelated changes in various signals appear differently. To distinguish between mild, moderate, and severe pain, a patient-centered approach has been developed [3-6].

Conclusion

Individual reactions of the patients were observed. Some of them showed a lot of emotion with their faces, while others showed almost no emotion at all. Despite the fact that some single modalities appear to adhere to the PRR level well, there may be a range of reliability in them. For example, when pain started, EDA changed very much, but the signal also shows how the brain responded to emotions. Consequently, it appears that the multimodal dataset and feature fusion are heavily utilized in PRR assessment. Aside from that, it's important to note that the majority of the experiments didn't use 10-fold cross-validation but rather one patient out. It demonstrates that physiotherapy patients are unable to adapt to pain. As a result, a subject-subordinate solution to this problem is desired. As a result, the patient-specific model will be developed in our subsequent work. The personalized model will likely make it possible to examine the subjective patient response in greater depth and increase efficiency. Additionally, in manual physiotherapy, attempting to identify more than three classes may perform better and provide a more accurate representation of the nature of pain.

Acknowledgement

None.

Conflict of Interest

The authors declare that there is no conflict of interest associated with this manuscript.

References

1. Kromer, Thilo O., Rob A. de Bie and Caroline HG Bastiaenen. "Effectiveness of individualized physiotherapy on pain and functioning compared to a standard exercise protocol in patients presenting with clinical signs of subacromial impingement syndrome. A randomized controlled trial."BMC Musculoskelet Disord11 (2010): 1-13.

Google Scholar, Crossref, Indexed at

2. Stecco, Antonio, Andrea Meneghini, Robert Stern and Carla Stecco, et al. "Ultrasonography in myofascial neck pain: Randomized clinical trial for diagnosis and follow-up."Surg Radiol Anat 36 (2014): 243-253.

Google Scholar, Crossref, Indexed at

3. Wideman, Timothy H., Robert R. Edwards, David M. Walton and Marc O. Martel, et al. "The multimodal assessment model of pain: A novel framework for further integrating the subjective pain experience within research and practice."Clin J Pain35 (2019): 212.

Google Scholar, Crossref, Indexed at

4. Williams, Amanda C. de C. and Kenneth D. Craig. "Updating the definition of pain."Pain157 (2016): 2420-2423.

Google Scholar, Crossref, Indexed at

5. Herr, Keela, Patrick J. Coyne, Elizabeth Ely and Céline Gélinas, et al. "Pain assessment in the patient unable to self-report: Clinical practice recommendations in support of the ASPMN 2019 position statement."Pain Manag Nurs 20 (2019): 404-417.

Google Scholar, Crossref, Indexed at

6. Oliveira, Juliana S., Cathie Sherrington, Elizabeth RY Zheng and Marcia Rodrigues Franco, et al. "Effect of interventions using physical activity trackers on physical activity in people aged 60 years and over: A systematic review and meta-analysis."Br J Sports Med 54 (2020): 1188-1194.

Google Scholar, Crossref, Indexed at