Abstract
Persistent pain is associated with twice the risk of developing frailty: in the U.S., over 36.0% of older adults ≥ 65 y.o. report chronic pain (CP) but more than 37% long-stay nursing home residents with persistent pain are under- or untreated. Communication barrier (e.g., Alzheimer’s Disease and Related Dementia) is one of the most significant sources of under-treatment of pain. In this work, we collect a small pain dataset on young and older subjects with chronic conditions, and we aim to develop an automated pain assessment model for timely detection of unexpressed pain and prevention of frailty using non-invasive, objective sensing. We do so by investigating physiology recovery dynamics, pain history, and other features related to frailty, and explicitly model such features to predict pain robustly. Our dataset is the first to collect pain on older adults with chronic conditions, and we hope it can contribute to the automated pain assessment and frailty prevention in non-communicative people.
