Clinician Perspectives on Automated Facial Analysis for Pain Assessment: A Cross-Sectional Survey of Physicians, Advanced Practice Providers, and Nurses

The study examined the views of 87 clinical professionals (physicians, advanced practice providers, and nurses) on automated facial analysis using artificial intelligence for pain assessment. Current pain assessment practices were also evaluated. Most clinical professionals relied on observed indicators including vital signs (88.5%), physical examination findings (85.1%), and facial expressions (79.3%) when assessing pain. Clinician attitudes toward automated pain assessment were neutral to positive. Half of respondents (50.6%) agreed that such a system would be useful for long-term patient monitoring. The majority of clinicians (86.5%) saw value in the system for non-communicative patients, for patients with dementia or cognitive impairment (78.4%), and for patients with language barriers (62.2%). Nearly 60% of clinicians required a minimum accuracy of 90% before routine use. Main concerns included alert fatigue and unequal system performance across different patient populations.