Automated Recognition Of Dentists’ Poor Postures Using Artificial Intelligence

This study aimed to create and test an artificial intelligence (AI)-based system for the automated evaluation of dentists' clinical sitting postures. The dentists' sitting postures were compiled into a dataset. A training-validation set comprised 80% of the dataset, while a test set made up 20%. Segmenting video frames, extracting keypoints, calculating geometric angles, and training and validating classification models were the main algorithmic components. Performance measurements utilized included accuracy, F1-score, recall, confusion matrix, and Kappa coefficient. The model's performance was assessed using a 5-fold cross-validation. The model exhibited outstanding performance on the test set, with an accuracy of 98.2%, an F1-score of 0.974, a sensitivity of 0.983, a specificity of 0.995, and an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.999. It accurately identified typical poor postures, such as forward head tilt, lateral tilt, and shoulder asymmetry. It surpassed the conventional threshold-based classification method (P < 0.01). This research has created and tested an AI-based automated clinical posture evaluation system. The technology facilitates quantitative assessment of dentists' bad sitting postures, and has the potential to offer a dependable instrument for early prevention and management of occupational musculoskeletal problems.