A Non-Invasive Simplified Model for Estimating Lower Limb Muscle Forces During Slow Gait in Older Adults and Post-Stroke Individuals

This study proposes a non-invasive, simplified muscle force estimation model (NSMFEM) designed for elderly individuals and stroke patients under slow walking conditions. The model estimates lower limb muscle forces dynamically using only kinematic parameters—with real-time muscle fiber length as the key variable—thus avoiding the limitations of traditional surface electromyography (sEMG)-based approaches such as environmental interference, signal noise, and difficulty in obtaining deep muscle sEMG. A personalized Digital Twin Musculoskeletal Model (DTMSM) was constructed by scaling a reference kinematic model and calibrating muscle origin/insertion markers based on individual anthropometry. Muscle architecture indices were derived from a multiple regression model with publicly available anatomical data. Twelve elderly subjects (eight healthy ESND and four post-stroke ESP) were evaluated at varying walking speeds. Results at slow speeds (X-slow and slow) show strong Pearson correlations between NSMFEM predictions and reference data for the majority of nine representative lower limb muscles (e.g., TFL, Iliacus, Pectineus, Tib_Ant, Soleus); passive forces of TFL, Iliacus, and Vas_Int also correlate strongly. As speed rises, correlations for some muscles (e.g., Vas_Int, Tib_Post) decline, reflecting the growing influence of segmental acceleration and muscle activation—factors omitted in the model. For stroke patient gait (ESP), Spearman analysis indicates maintained strong correlations for affected side muscles Glut_Max1, TFL, Pectineus, and Soleus, supporting the model’s utility in stroke rehabilitation assessment. Overall, NSMFEM offers a practical, sEMG free method for non-invasive dynamic muscle force estimation in slow walking elderly and post-stroke populations, aiding functional assessment and personalized rehabilitation planning. Future efforts will aim to incorporate muscle activation corrections to extend the model to faster walking speeds.