The Impact of Human Posture on Prevention of Neck Musculoskeletal Disorders: A Multivariate Time-Series Prediction Approach Using LSTM

Introduction: Bending is an essential human movement that requires precise coordination between muscles and organs for optimal performance. “Text Neck Syndrome” refers to pain resulting from poor neck posture when looking down at a phone for long periods. Many studies link forward head posture or neck curvature to various tasks, particularly while using mobile phones and computers. The weight exerted on the neck increases significantly as the degree of curvature increases. However, this increase has not been adequately correlated with the duration of neck curvature, which is a critical factor influencing changes in neck weight over time and its negative health impacts, especially on the cervical vertebrae. Aim: In this paper, two distinct solutions are proposed. Methods: The first is a mathematically derived model that considers factors such as head weight and bending duration as critical determinants affecting the efficiency of neck movement. This model analyzes the relationship between skull weight, bending angle, and time to achieve maximum motor efficiency. The second solution employs a multivariate time-series prediction approach using an LSTM model based on the multiple_IMU dataset. This method determines the extent of the problem and provides alerts for the allowable duration of neck bending. The model incorporates factors such as skull size, weight-to-height ratio, and other parameters to evaluate the impact. Results: the results demonstrate high accuracy and robust performance, with the LSTM model achieving a test loss of 0.00046, a test mean absolute error (MAE) of 0.00925, an accuracy of 98.93%, and an F1 score of 0.9935, which is higher than that of other Deep Learning models such as CNN and GRU. Conclusions: These findings underscore the potential of mathematical modeling and AI-based techniques in addressing the health challenges posed by prolonged poor neck posture.