Assessing procedural pain in infants: a feasibility study evaluating a point-of-care mobile solution based on automated facial analysis

The management of procedural pain in infants is suboptimal, in part, compounded by the scarcity of a simple, accurate, and reliable method of assessing such pain. In this study, we aimed to evaluate the psychometric properties of the PainChek Infant, a point-of-care mobile application that uses automated facial evaluation and analysis in the assessment of procedural pain in infants.Video recordings of 40 infants were randomly chosen from a purposely assembled digital library of 410 children undergoing immunisation as part of their standard care in Prishtina, Kosovo, between April 4, 2017, and July 11, 2018. For each infant recording, four 10 s video segments were extracted, corresponding to baseline, vaccine preparation, during vaccination, and recovery. Four trained assessors did pain assessments on the video segments of 30 infants, using PainChek Infant standard, PainChek Infant adaptive, the Neonatal Facial Coding System-Revised (NFCS-R) single, the NFCS-R multiple, and the Observer administered Visual Analogue Scale (ObsVAS), on two separate occasions. PainChek Infant's performance was compared to NFCS-R and ObsVAS using correlation in changes in pain scores, intra-rater and inter-rater reliability, and internal consistency.4303 pain assessments were completed in two separate testing sessions, on Aug 31, and Oct 19, 2020. The study involved videos of 40 infants aged 2·2-6·9 months (median age 3·4 months [IQR 2·3-4·5]). All pain assessment tools showed significant changes in the recorded pain scores across the four video segments (p≤0·0006). All tools were found to be responsive to procedure-induced pain, with the degree of change in pain scores not influenced by pre-vaccination pain levels. PainChek Infant pain scores showed good correlation with NFCS-R and ObsVAS scores (r=0·82-0·88; p<0·0001). PainChek Infant also showed good to excellent inter-rater reliability (ICC=0·81-0·97, p<0·001) and high levels of internal consistency (α=0·82-0·97).PainChek Infant's use of automated facial expression analysis could offer a valid and reliable means of assessing and monitoring procedural pain in infants. Its clinical utility in clinical practice requires further research.PainChek.