Classifying office workers with and without cervicogenic headache or neck and shoulder pain using posture-based deep...

Area of Science:

• Biomedical Engineering
• Computer Science
• Musculoskeletal Health

Background:

• Office workers frequently experience cervicogenic headache (CH) and neck/shoulder pain (NSP).
• Habitual sitting posture is a potential contributing factor to these conditions.
• Objective methods for identifying postural risks are needed.

Purpose of the Study:

• To develop and evaluate deep learning models for classifying office workers with and without CH and/or NSP.
• To analyze habitual sitting posture images for distinct patterns associated with these conditions.

Main Methods:

• A multicenter, retrospective study analyzed 904 sitting posture images from 531 office workers.
• Three deep learning models (VGG19, ResNet50, EfficientNet B5) were trained to classify CH, NSP, and combined CH+NSP.
• Performance was assessed using AUC, accuracy, sensitivity, specificity, and F1 score, with Class Activation Mapping (CAM) for visualization.

Main Results:

• ResNet50 achieved the highest performance for CH classification (AUC 0.782, accuracy 0.750).
• NSP classification yielded modest results (ResNet50 accuracy 0.677).
• EfficientNet B5 showed the highest AUC for combined CH+NSP classification (0.744). CAM identified condition-specific postural focus areas.

Conclusions:

• Deep learning models demonstrate potential for classifying CH and NSP based on sitting posture.
• These models can detect subtle postural patterns indicative of musculoskeletal conditions, suggesting applications in early detection.
• A multimodal assessment approach remains crucial due to the complex link between posture and pain.