Pre-clinical measurements of internal pain states have traditionally depended on the manual scoring of reflexive paw withdrawal in response to noxious stimuli. This approach, however, struggles to differentiate whether the withdrawal is actually motivated by pain, as much of the rapid somatosensory stimuli motivated motor response happens on millisecond timescales that are not detectable by the human eye.
By combining high speed videography, automated paw tracking, and machine and deep learning methods, Jones and colleagues created the Pain Assessment at Withdrawal Speeds (PAWS) statistical software platform: an R package that takes in paw trajectory data from pose tracking algorithms such as SLEAP, and provides an automated scoring of pain.
The platform extracts seven features from a univariate projection of the position of the paw that is then combined into a unified pain score. The group has validated the platform, identifying a hypersensitive mouse strain, and showed precision of the platform by successfully detecting chemogenetically induced hypersensitivity via basolateral amygdala pain ensemble activation. Overall, PAWS provides a high throughput and user-friendly pain assessment platform for making more objective and robust pain measurements in pre-clinical research.