Lateral turning is a critical aspect of nursing care for individuals with disabilities, and the autonomous execution of the lateral turning process by devices has become one of the key tasks in the development of unmanned nursing care. To enhance the safety and intelligence of lateral turning devices, pressure information, which is inevitably generated during the human lateral turning process, is utilized as a core indicator to construct a model for estimating the human lateral turning state and guiding the control system′s execution. Analytical mechanics is employed to construct a matrix function from seven pressure points at the shoulders and hips during the lateral turning process. Based on anatomical principles, height and weight are incorporated as variable parameters to achieve active adaptation of the model. Cosine similarity and Pearson correlation coefficient are employed to jointly assess actual and theoretical pressure, yielding a minimum similarity of 0.826 9, thereby improving the model′s robustness and enabling human rollover state estimation. The constructed human lateral turning state estimation model further analyzes the lateral turning movements of individuals with disabilities, which holds significant implications for the intelligentization of rehabilitation aids, health status assessment, and routine home care.