Background and Significance
Pain is a subjective distressing experience associated with actual or potential tissue damage with sensory, emotional, cognitive and social components. Historically, pain in humans has been measured using subjective self-report scales to determine presence and severity. While this can be useful information, self-report is a problematic metric for both diagnostic and research purposes. For example, self-report is impossible to obtain in various clinical populations, such as unconscious patients or patients with cognitive impairments who have difficulties in verbal or motor expressions. Further, comparisons between people reporting their pain is difficult to do with confidence, as these self-report metrics are highly subjective, depend on previous history with pain and other cognitive and behavioral factors, and can vary over time. Therefore, while current assessment of pain largely relies on the self-report of an individual, the development of an objective, automatic detection/measure of pain may be useful in many research and clinical applications. Such approaches, if successful may not only detect pain, but may provide for a more rational therapeutic intervention. Hence, the objective of this work is to evaluate the use of physiological and behavioral metrics as markers of pain, and to develop automatic methods for objectively quantifying pain.
To achieve the aforementioned objective, we propose to study pain responses in healthy adults subjected to experimental pain. In this work, we specifically propose to explore the use of (a) autonomic parameters reflecting sympathetic and parasympathetic nervous activity (viz., skin conductance and heart rate variability), (b) brain hemodynamic changes as captured by functional near-infrared spectroscopy, and (c) facial expressions from video as markers for the presence of acute pain. In Aim 1, we will characterize changes in these signals in response to pain. In Aim 2, we will develop automatic machine learning approaches for the detection of the presence of pain and the inference of pain intensity levels. In Aim 3, we will extend these approaches by introducing personalized machine learning models that can account for individual differences in pain responses while learning from data from across the population. Finally, in Aim 4, we will evaluate the performance and reliability of the proposed algorithms for different use cases, and discuss potential practical implications of their implementation.
To accomplish the aims of this research, we will (a) use publicly available datasets and (b) collect new data from healthy volunteers subjected to experimental pain. The resulting data will include pain due to thermal (heat), mechanical and electrical stimulation. While no single dataset will contain all signal modalities considered simultaneously, different combinations of these will be available. Therefore, we will explore both uni-modal and multi-modal approaches to pain detection. Multiple studies have already characterized stereotypical physiological and behavioral responses to pain, and developed automatic methods for the detection of the presence of pain. We will extend this work by developing novel algorithms for the inference of pain intensity. There is often great variability in how people perceive, experience, and physiologically and behaviorally express pain, hence stemming efforts to build a one-size-fits-all pain recognition system. To address this, we will propose novel state-of-the-art machine learning methods for the personalization of the inference process. This approach will result in models tailored specifically for individuals that still account for the broader population data.