Few objective pain assessment techniques are currently available for use in clinical settings. Clinicians typically use subjective pain scales for pain assessment and management, which has resulted in suboptimal treatment plans, delayed responses to patient needs, over-prescription of opioids, and drug-seeking behavior among patients. This project will investigate science-based methods to build a robust Continuous Objective Multimodal Pain Assessment Sensing System (COMPASS) and a clinical interface capable of generating objective measurements of pain from multimodal physiological signals and facial expressions. COMPASS will allow objective measurements that can be used to significantly improve pain assessment, pain management strategies, reduce opioid dependency, and advance the field of pain-related research. The educational plan will include activities to engage patient training, K-12 students, minorities and underrepresented groups, as well as general public. These outcomes will also lead to development of a diverse work force needed to support advanced medical technologies and services.
Using advanced biosensing systems, data fusion algorithms and machine learning models, this project will develop a robust, reliable, and accurate pain intensity classification system, COMPASS, for estimating pain intensity experienced by patients in real-time on a 0-10 scale, which is the standard scale used by physicians in clinical settings. In the initial phase of the project, the team will conduct a pilot at Brigham and Women's Hospital to experiment with the different elements for developing the sensing systems and collect data to develop data fusion algorithms and machine learning models. In the later phase of the project, the team will collect an extensive set of data to train and validate the fully implemented COMPASS. Physiological sensor data from electroencephalograph, facial-expression, patient self-reported pain scales, and physician/nurse assessed pain scales will be collected from the subjects as they experience pain modulated by medical therapies that cause patients pain. The project will investigate evidence-based machine learning and feature extraction methods for physiological signals and facial-expression images. This highly interdisciplinary research will make significant contributions to the areas of pain assessment and management, human factors and patient safety.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
