Pain is a distressing feeling often caused by intense or damaging stimuli and commonly defined as an unpleasant sensory and emotional experience associated with actual or potential tissue damage. In clinical medicine, pain is often regarded as a symptom of an underlying condition. Since pain is a complex and subjective phenomenon, objectively defining pain has been difficult. This project is designed to develop machine learning and computer algorithms for enhanced and objective pain recognition based on unique behavioral and neuronal activity patterns recorded in a novel animal pain model. The investigators will then apply and test the algorithmic tool to two local and systemic inflammatory conditions that are often associated with severe pain. Successful completion of this project will potentially lead to development of innovative pain assessment tools for both mechanistic exploration and future management of pain.

Technically, the project has two specific aims. The first one is to establish machine learning-based pain recognition algorithms built on unique behavioral and central neuronal activity patterns induced by pain. The investigative team has established a novel animal model of neuropathic pain (trigeminal neuralgia) that produces unique behavioral (e.g., orbital tightening and face grimace) and central neuronal activity patterns as documented by video recording and in vivo two-photon imaging, respectively. This model has generated wealth of high-quality data sets directly related to spontaneous pain and offers a rich source as training materials for machine learning using convolutional neural network, which will enable the team to develop computer algorithms for enhanced pain recognition. The second aim is to examine the pain recognition algorithms developed in Aim 1 by investigating its performance in animal models with clinical relevance. Specifically, the team will employ ankle joint arthritis and bacterial sepsis as models for local and systemic inflammatory conditions, respectively. Using conventional methods, spontaneous pain behaviors have not been reliably captured in these models. The team plans to prove that pain recognition algorithms based on convolutional neural network established in Aim 1 are powerful and versatile to reveal novel patterns indicative of spontaneous pain, despite vastly different underlying mechanisms of pain. The results will be highly valuable to inform future studies aiming at optimizing treatment regimens and improving long term outcomes of the diseases.

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.

Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Maryland Baltimore
United States
Zip Code