A research-stage machine learning framework has demonstrated the feasibility of screening adolescents for forward head posture using side-view photographs and tabular classification. In a large-scale population study involving 1,577 Chinese adolescents recruited for posture assessment, researchers combined automated computer vision landmark detection with machine learning to identify the common postural abnormality.
Forward head posture is a widespread postural abnormality among adolescents that is linked to musculoskeletal imbalances and potential long-term health risks. To evaluate and screen for the condition non-invasively, the investigative team developed an end-to-end framework integrating human pose estimation with tabular machine learning models.
The workflow applied RTMPose to extract key anatomical landmarks from side-view images, which were then used to computationally derive seven predefined, anatomy-informed posture features. A TabPFN classifier trained on these extracted features achieved an F1 score of 82.46% and an AUC of 0.837.
To interpret the model, a permutation-based SHAP analysis identified the horizontal angle between the right eye and shoulder as the strongest predictive contributor. A SHAP dependence plot also suggested a potential nonlinear dependence between features beyond pairwise correlation. However, these associations solely describe the behavior of the fitted model and do not establish causal biomechanical effects.
These findings support the feasibility of a non-invasive and interpretable image-based framework for screening. Nevertheless, the framework remains a research-stage classifier, meaning prospective clinical validation and implementation evaluation are required before any practical use in school-based or public-health settings.
Subject of Research: Machine learning screening for adolescent forward head posture
Article Title: From keypoints to classifications: A SHAP-driven TabPFN framework for forward head posture screening
Article References: Liang, S., & Wang, Z. (2026). From keypoints to classifications: A SHAP-driven TabPFN framework for forward head posture screening. PLOS One, 21(11), e0362855. https://doi.org/10.1371/journal.pone.0362855
Image Credits: AI Generated
DOI: 10.1371/journal.pone.0362855
Keywords: forward head posture, postural abnormality, adolescents, RTMPose, TabPFN, human pose estimation, SHAP analysis, tabular machine learning, side-view images, anatomical landmarks, posture screening, musculoskeletal health, model interpretability, feature importance, F1 score, AUC, predictive modeling
Cite Scienmag News
Teresa Odom. (November 21, 2026). Machine Learning Model Screens Adolescent Forward Head Posture From Images. Scienmag. https://scienmag.com/machine-learning-model-screens-adolescent-forward-head-posture-from-images/
Teresa Odom. "Machine Learning Model Screens Adolescent Forward Head Posture From Images." Scienmag, 21 November 2026, https://scienmag.com/machine-learning-model-screens-adolescent-forward-head-posture-from-images/. Accessed 21 November 2026.
Teresa Odom. "Machine Learning Model Screens Adolescent Forward Head Posture From Images." Scienmag. November 21, 2026. https://scienmag.com/machine-learning-model-screens-adolescent-forward-head-posture-from-images/








