Enhancing pain assessment in cognitively impaired residents using AI-enabled technology: a case study in long-term care

Long-term care (LTC) residents are at high risk for underrecognized/untreated pain, particularly when they are unable to self-report. This quality improvement project presents the implementation of a commercially available artificial intelligence (AI)-powered application for pain assessment (PainChek ® ) in a Canadian LTC home. Specifically, it: (1) evaluates nurses’ confidence in assessing residents’ pain and current practices; (2) assesses factors influencing technology adoption; and (3) investigates satisfaction, use, and impacts of the technology. PainChek ® was implemented across four units (160 beds) following 18 training sessions (evaluation period: June-October 2024). Using a pre-post design, surveys were administered to nurses ( n = 65; [1–5 Likert scale]) before (T1) and 12 weeks post-implementation (T2). The surveys measured their confidence in pain assessment, adoption factors, perceived impacts, use and satisfaction. Secondary data on technology use and residents’ pain scores were extracted from PainChek ® data and analyzed in SPSS v29. Nurses reported confidence in assessing residents’ pain, strong support for the new technology, and agreement that it aligned with existing workflows (medians ≥ 4). Post-implementation, nurses rated the technology as easy to use and reported high satisfaction, frequent use and encouragement from colleagues and supervisors. Perceived impacts on resident care were positive (median = 4), although nurses were less certain that PainChek ® reduced responsive behaviours (median = 3). No significant changes were observed in nurses’ confidence in their own pain assessment abilities, existing pain assessment practices, and perceived impacts on workflow. However, nurses' confidence in the reliability of the technology slightly decreased at T2 ( p = 0.009). The most frequently reported barriers included limited tablet availability (20%) and the time required to complete assessments (18%). Overall, 3,826 assessments were completed using PainChek ® , identifying mild-to-severe pain in 764 cases (20%). Technology use improved over time: the proportion of follow-up assessments increased from 69% in month 1 to 78% in month 3, while the proportion of assessments recording a pain score of 0 decreased from 14% to 6.2%. AI-enabled technology may support improved pain assessment with minimal workflow disruption. Fewer assessments following PainChek ® implementation may reflect more appropriate/precise methods, enabling improved pain detection and intervention. These findings provide preliminary insights into pain assessment practices in LTC homes and may inform future efforts to improve residents’ pain assessment.