Evaluation of Digital Face Recognition Technology for Pain Assessment in Young Children.

Objectives: Accurate assessment of pain in young children is challenging. An Emotion Application Programing Interface (API) can analyze and report 8 emotions from facial images. Each emotion ranges between 0 (no correlation) to 1 (greatest correlation). We evaluated correlation between the Emotion API with the FLACC scale (face, lets, activity, cry, and consolability) among children younger than 6 years old during blood sampling.

Methods: Prospective pilot exploratory study in children during blood sampling. Pictures with facial expressions were uploaded to Emotion API program. Primary outcome was the correlation coefficient between FLACC scale and emotions. Secondary outcomes included maximal correlation of each emotion for 3 pictures-before, during and after needle penetration; and the average of each emotion for 9 pictures-4 before, 1 during and 4 after needle penetration to the skin.

Results: A total of 77 children were included. During needle penetration, SADNESS was significantly correlated (0.887, P<0.05), and NEUTRAL was negative correlated with the FLACC scale (-0.841; P<0.05). The maximal correlation of each emotion showed increase in SADNESS and decrease in NEUTRAL emotions during, compared to before, needle penetration. Similar findings were observed when the average of each emotion was compared during to before needle penetration.

Discussion: During a blood test procedure, young children show higher SADNESS and lower NEUTRAL emotions as reported by the Emotion API. This software program may be useful in reporting emotions related to pain in young children, and more research is needed to compare its validity, reliability and real-time application compared to the FLACC scale.