Frontiers in Public Health (Oct 2015)
Building An Intelligent Wearable Movement Tracking Device to address Psychological Barriers to Mobility in Chronic Musculoskeletal Pain
Abstract
BACKGROUND Movement is an important aspect of life: it enables social interactions, household activities, employment, and physical fitness. In chronic musculoskeletal pain (CMP)–a prevalent medical condition where pain persists in the absence of tissue damage–wrong association of pain with movement and the accompanying fear of (re)injury lends to guarding or movement avoidance [2-4]. Mobility in CMP is reduced as a result of distress due to pain and reduced confidence in the ability to perform physical activity [8]. Addressing these psychological barriers enables improved mobility and, thus, higher quality of life [7-8]. Physiotherapists, for example, grade exercise difficulty and feedback in physical therapy to the level of confidence a CMP patient has in performing the exercise [7]. However, CMP is largely self-managed (i.e. with limited support from medical professionals) and people with CMP are often ill-equipped to constructively adapt movement in everyday routine to their pain-related psychological states [8]. We investigate the feasibility of affect-aware technology to address this problem. AIM The aim of our work is to design technology that equips the user to manage psychological barriers to functional physical activity. For such technology, the ability to automatically monitor pain-related psychological states is necessary. It also needs to be viable for use in mobile settings. Thus, our work has three main objectives: • to investigate the automatic recognition of pain-related psychological states. We focus on levels of pain, distress, and movement confidence. • to explore the feasibility of wearable technology for the afore-mentioned objective. • to examine the possibility of affect-aware functionality in personalizing feedback during physical activity. It is important to address pain, distress, and movement confidence in CMP as they contribute to reduced mobility and quality of life. They are particularly significant as monitoring them will enable (i) physical activity to be tailored on the fly. e.g. the technology may suggest that the user take a break when having increasing pain while performing an activity (this is important so as not to reinforce wrong associations of pain with movement), (ii) personalized feedback such as an encouraging message when the user has low movement confidence, and (iii) self-awareness, for example, progress may be quantified to the user as weekly visual representations of their levels and quality of physical activity with distress accounted for (this adapted definition of progress is necessary due to the nature of CMP [7]). METHODS AND RESULTS Literature in CMP and affective computing point to the efficacy of body movement information in discriminating between levels of psychological states. This motivates our use of kinematics and muscle activity as modalities in building our recognition systems. The data we use is from the Emo-Pain corpus [1], which consists of motion capture and surface electromyography data from people with chronic low back pain and healthy control subjects performing functional movements. In our work, we investigate automatic recognition in three movement types–forward reach (as if to reach across a table), bending to the floor (as in bending to pick up a bag from the floor), and sit-to-stand. In our work, we have so far developed systems that can differentiate between healthy control subjects, lower level pain in CMP, and higher level pain in CMP with accuracies of 86%, 94%, and 80% in the three movement types respectively [5-6]. We developed separate systems for the different movements as the context of the movement is important in the modelling of pain levels. We are, furthermore, exploring the possibility of a wearable recognition system. To enable this, we investigate the minimization of the number of anatomical segments that need to be tracked to recognize the psychological states. In the case of pain level recognition, we found that collectively for the three movement types considered, the movement of the head, torso, left arm, and left and right thighs and legs need to be tracked. This is in addition to the tracking of the upper trapezius and lumbar paraspinal muscle groups. We are currently developing a wearable prototype that uses low-cost gyroscopes and accelerometers for motion capture and surface electromyography for muscle activity tracking. An off-the-shelf smartphone is used to control units of the prototype and collate the data from them. This prototype allows us to collect kinematic and muscle activity data to build the affect-monitoring functionality into it. In addition, it enables contextual inquiry of people with CMP about how such functionality may be designed to be useful to them in everyday physical activity–our design would also take into consideration the view of appropriate medical professionals (e.g. physiotherapists) as in the requirements gathered in [7]. Figure 1 shows a person with low back pain wearing the current iteration of the prototype on the head, arm, trunk, and upper and lower legs. CONCLUSION Our studies so far promise the possibility of developing affect-aware wearable technology that can equip people with CMP to overcome psychological barriers to physical activity. REFERENCES [1] Aung, M. et al. (2015). The automatic detection of chronic pain-related expression: requirements, challenges and a multimodal dataset. IEEE Trans in Affective Computing. [2] Breivik, H. et al. (2006). Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment. European J Pain, vol. 10(4), pp. 287–333. [3] Crombez, G. et al. (1999). Pain-related fear is more disabling than pain itself: evidence on the role of pain-related fear in chronic back pain disability. Pain 80(1), 329-339. [4] International Association for the Study of Pain. (1986). Introduction. Pain, vol. 24(Supplement 1), S3–S8. [5] Olugbade, T., Aung, M., Marquardt, N., de C. Williams, A., and Bianchi-Berthouze, N. (2014). Bi-modal detection of painful reaching for chronic pain rehabilitation systems. ICMI, pp. 455-458. [6] Olugbade, T., Bianchi-Berthouze, N., Marquardt, N., and de C Williams, A. (2015). Pain level recognition using kinematics and muscle activity for physical rehabilitation in chronic pain. ACII. [7] Singh, A. et al. (2014). Motivating people with chronic pain to do physical activity: opportunities for technology design. CHI, p. 2803. [8] Turk, D. and Okifuji, A. (2002). Psychological factors in chronic pain: evolution and revolution. J Cons & Clin Psychol, vol. 70, p. 678.
Keywords
