Research in Automatic Pain Assessment

This chapter introduces key concepts of nociception and pain. Therefore, it delves into current research on automatic pain assessment (APA), exploring a variety of techniques that aim to objectively measure pain through behavioral-based approaches such as speech analysis, and facial expression recognition, and neurophysiology-based pain detection methods such as electrodermal activity, electroencephalography, electrocardiography, photoplethysmography, functional magnetic resonance imaging, and other approaches. For each technique, it reviews the latest advancements in machine learning models used for pain detection and the challenges associated with assessing pain. Additionally, this chapter also discusses the limitations of current APA technologies, such as variability in individual pain responses and the influence of external factors. On these bases, this chapter offers perspectives and proposes multimodal frameworks for continuous and non-invasive pain monitoring, such as the “Pain Holter” concept. It concludes by emphasizing the need for rigorous validation studies and the future potential of these tools to support more personalized and accurate pain management.