MuReB-Net: Advanced Residual Network with Modified Residual Blocks for Automatic Pain Level Classification

Pain is considered one of the most critical alerts for humans, signaling potential health issues that require immediate attention. Consistently monitoring and accurately recognizing patient pain remains a significant challenge in healthcare systems. Automated pain detection is seen as a promising solution to this challenge. This paper proposes MuReB-Net (Multiple Modified Residual Block Network), a deep network architecture featuring multiple modified residual blocks, using the advantages of different modules and attention mechanisms to detect pain effectively. The model’s performance is evaluated using the UNBC-McMaster Shoulder Pain Expression Archive Database. The results demonstrate the capability of MuReB-Net to accurately determine the presence of pain and classify pain into four distinct levels. The results indicate an accuracy of $96.55 \%$ for the two-level pain classification (pain vs. no pain) and $\mathbf{9 1 . 4 \%}$ for the four-level pain intensity classification.