Facial Expression-Based Pain Analysis Using Deep Learning Models
Keywords:
RNN, MIntPAIN, BioVid, UNBC-McMaster, Deep LearningAbstract
Pain is a key physiological signal of potential harm or dysfunction. While many can describe their pain, others, such as infants, unconscious patients, or those with speech impairments, cannot. In these cases, automated pain assessment is essential. Facial expressions offer a reliable, objective, and noninvasive means for real-time pain analysis. In This paper reviews deep learning algorithms for facial expression-based pain detection, comparing models developed between 2017 and 2025. It examines performance across datasets such as UNBC-McMaster, BioVid, MIntPAIN, and paediatric pain datasets. Advanced architectures, including convolutional neural networks, recurrent neural networks, Vision Transformers, and hybrid ensembles, have achieved up to 98.7% accuracy. The paper concludes with an overview of current challenges and future directions for real-time clinical use.
References
Runwag,KexuFeng,Weigh,”Hybrid RNN-ANN based deep Physiological Network for pain Recognition”,AnnyIntconf ,IEEE med Bio soc,2020, page no. 5584-5587.
Raul Fernqdez,RaisRomero,”Pain assessment based on FNIRS Using Bidirectional LSTM “,arxivcs,24 dec 2020.
Ming Bishay,GrahamPage,MohamaadMavadati,” PainNet: statistical relation network with Episode–based training for pain estimation”, Arxiv,8 April 2025.
Jingzhou,Xiaopena,Feish,” Recurrent Conventional Neural network regression for continue pain intensity estimation in video”,3 May 2010.
Ehsan Othman, Phillppwerner, Frerksaxena,” An automatic system for continuous pain intensity monitoring based on analyzing data from uni and bidirectional and multimodal”, Sensor2022, 15 July 2022, pg 210-222.
Ghazal Bargshady ,Xujuan Zhou, RavineshCdeo,” Ensemble neural network approach detectionpain intensity from facial Exprisson “,Pubmed,7 september 2020,DOI :10.1016 ,artmed 2020.
Ji-Tuozhang,Xiao-Yihu,WenduanPubmed ,“ application of deeplearning based facial pain recognition model for post operative pain assment”,13 june2025,DIO:10.1016/Jjeline.2025.111898.
MohamaadTavakoilan,AbdenoulHodid,” Spatiotemporal convolutional Neural network for automatic pain intensity Estimation from facial Dynamics “,Innternational journal of Computer vision ,2019.
ShrivastavaAarti,BhadouraiyaSanajy,”Efficient Pain Recognition via Deep Learning: A Comparative Analysis of Lightweight and Deep CNN Models”. (2025). Ianna Journal of Interdisciplinary Studies,ISSN(O):2735-9891,ISSN(P):2735-9883, 7(2), 42-61.
AI Summer, "Best deep CNN architectures and their principles: from AlexNet to EfficientNet", [Online]. Available: https://theaisummer.com/cnn-architectures/. [Accessed: Sep. 9, 2025].
Downloads
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
