A Systematic Literature Review of Methods and Technologies for Detecting Pain in Infants Through Facial Expression Analysis

Abstract

Pain assessment in infants, particularly neonates, remains a challenging task due to their inability to verbally communicate pain. Therefore, pain evaluation commonly relies on indirect indicators, including facial expressions, crying behavior, and physiological responses. This study aims to review existing methods and technologies for infant pain detection through facial expression analysis using a Systematic Literature Review (SLR) approach based on the PRISMA framework. The review process involved several stages, including identification, screening, eligibility assessment, and inclusion of relevant studies, resulting in 27 articles selected for further analysis. The findings indicate that Deep Learning-based approaches, particularly Convolutional Neural Networks (CNNs), are the most frequently applied techniques for analyzing infant facial expressions in pain detection systems. Furthermore, recent studies have explored multimodal approaches by combining facial expressions with crying sounds and physiological signals to improve detection performance. Despite these advancements, several challenges remain, including limited data availability, lack of standardized public datasets, and difficulties in validating models across different clinical settings. Overall, artificial intelligence-based approaches demonstrate significant potential in supporting more objective and consistent infant pain assessment. This review provides insights into current research trends and future directions for developing reliable pain detection systems to assist healthcare professionals in neonatal care.

Keywords

Infant Pain; Facial Expression Analysis; Deep Learning; Machine Learning; Systematic Literature Review