Close
The EmoPain dataset is a semi-ecological multimodal dataset of people living with chronic pain and healthy participants engaged in everyday movement used in chronic pain physical rehabilitation. It contains body movement data (IMU sensor suit and EMG sensors) and facial together with variety of labels related to the pain and emotional experience of the person as well as the type of movement the person is performing. An extension of the EmoPain dataset is the EmoPain@Home dataset.
The EmoPain dataset is part of the multidisciplinary EPSRC Emo&Pain project Body movement part of the dataset: (if interested, contact: Nadia Berthouze, n.berthouze@ucl.ac.k)
This part of the dataset contains movement data collected from 18 IMUs of 12 healthy and 18 CP participants. The placement of IMUs is illustrated in Fig 3(a). Four wireless surface electromyographic sensors (sEMG) were also used and placed on upper and lower sections on the back of a participant to capture the muscle activity.
In this paper, we focus on the movement data and leave the exploration of muscle activity to future work. As part of the EmoPain dataset, the annotation of protective behaviour was provided by four domain-expert raters, including 2 physiotherapists and 2 clinical psychologists. Each expert rater independently inspected the on-site video of each CP participant that was collected in synchrony with the wearable sensor data. They marked the timesteps where each period of protective behavior started and ended. The healthy participants were assumed to show no protective behaviour despite having their own idiosyncrasies.
In addition, protective behaviour and self-efficacy labels by 30 physiotherapists are also provided at movement-instance levels.
Additional resources
Dataset paper
M. S. H. Aung et al., The Automatic Detection of Chronic Pain-Related Expression: Requirements, Challenges and the Multimodal EmoPain Dataset. In IEEE Transactions on Affective Computing, vol. 7, no. 4, pp. 435-451, 1 Oct-Dec 2016, doi: 10.1109/TAFFC.2015.2462830.
Related papers
Chongyang Wang, Yuan Gao, Akhil Mathur, Amanda C. De C. Williams, Nicholas D. Lane, and Nadia Bianchi-Berthouze. 2021. Leveraging Activity Recognition to Enable Protective Behavior Detection in Continuous Data. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 5, 2, Article 81 (June 2021), 27 pages. https://doi.org/10.1145/3463508
Chongyang Wang, Temitayo A. Olugbade, Akhil Mathur, Amanda C. DE C. Williams, Nicholas D. Lane, and Nadia Bianchi-Berthouze. 2021. Chronic Pain Protective Behavior Detection with Deep Learning. ACM Trans. Comput. Healthcare 2, 3, Article 23 (July 2021), 24 pages. https://doi.org/10.1145/3449068
T. A. Olugbade, N. Bianchi-Berthouze, N. Marquardt and A. C. de C. Williams. Human Observer and Automatic Assessment of Movement Related Self-Efficacy in Chronic Pain: From Exercise to Functional Activity. In IEEE Transactions on Affective Computing, vol. 11, no. 2, pp. 214-229, 1 April-June 2020, doi: 10.1109/TAFFC.2018.2798576.
Temitayo A. Olugbade, Aneesha Singh, Nadia Bianchi-Berthouze, Nicolai Marquardt, Min S. H. Aung, and Amanda C. De C. Williams. 2019. How Can Affect Be Detected and Represented in Technological Support for Physical Rehabilitation? ACM Trans. Comput.-Hum. Interact. 26, 1, Article 1 (February 2019), 29 pages. https://doi.org/10.1145/3299095