Search for a command to run...
Qiaojia Zhou,1 Xuanni Huang,1 Yuanhong Lv,1 Zhitian Xiao,1 Shuli Luo,1 Zhiyong Wang,2 Yuan Li,3,4 Yingxin Li,3,4 Qiong Chen,3,4 Zhangbin Yu,5 Queyun Zhou1 1Department of Neonatology, Shenzhen Children’s Hospital, Shenzhen, People’s Republic of China; 2School of Biomedical Engineering, Harbin Institute of Technology, Shenzhen, People’s Republic of China; 3Department of Neonatal Nursing, West China Second University Hospital, Sichuan University, Chengdu, People’s Republic of China; 4Key Laboratory of Birth Defects and Related Diseases of Women and Children (Sichuan University), Ministry of Education, Chengdu, People’s Republic of China; 5Department of Neonatology, Shenzhen People’s Hospital, Shenzhen, People’s Republic of ChinaCorrespondence: Queyun Zhou, Department of Neonatology, Shenzhen Children’s Hospital, No. 7019 Yitian Road, Futian District, Shenzhen, Guangdong, 518038, People’s Republic of China, Email 18938690108@189.cn Zhangbin Yu, Department of Neonatology, Shenzhen People’s Hospital, No. 1017 Dongmen North Road, Luohu District, Shenzhen, Guangdong, 518020, People’s Republic of China, Email zhangbinyu@njmu.edu.cnObjective: Effective neonatal pain assessment is crucial for optimal analgesic management; however, it remains challenging, as traditional pain scales and single-modal intelligent assessment approaches continue to face substantial methodological and clinical limitations. Intelligent systems integrating multimodal data offer promising alternatives for enhancing objectivity and continuity of assessment. This systematic review aimed to identify, evaluate, and synthesize current intelligent neonatal pain assessment methods based on multimodal data fusion.Methods: Two investigators independently searched PubMed, Embase, Cochrane Library, and Web of Science databases for relevant studies published up to September 12, 2025. Studies reporting the use of a multimodal approach to neonatal pain assessment were included. The methodological quality of the selected studies was independently assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2. A systematic review of the findings was then conducted.Results: Nine studies met the inclusion criteria and were included in this review. Seven studies compared the accuracy of single-modal versus multimodal assessments of neonatal pain, with five also reporting the Area Under the Curve (AUC) values. All studies demonstrated that multimodal approaches achieved higher accuracy than single-modal methods. Specifically, 77.78% (7/9) of the studies successfully assessed pain intensity and distinguished between neonates with and without pain. However, only 44.44% (4/9) addressed methods for handling missing data, and merely 33.33% (3/9) utilized external validation sets.Conclusion: Multimodal pain assessment demonstrates superior accuracy in neonatal pain evaluation. However, the heterogeneity in modalities, outcome indicators, and performance statistics across existing studies limits the ability to identify an optimal combination of modalities. Further research focusing on standardization, clinical applicability, and robust validation is required to strengthen the evidence base in this field and facilitate clinical translation.Keywords: pain, multimodal, artificial intelligence, neonates, systematic review