Stroke rehabilitation is a critical area of research, with soft actuators playing an
increasingly important role in improving function recovery. These devices, particularly soft
finger actuators, are vital for enabling precise motion control in rehabilitation training
gloves, which are essential for restoring hand functionality. This study presents an in-depth
analysis of soft finger actuators using finite element modeling to evaluate their performance
across various structural configurations. By conducting detailed numerical simulations, the
research explores how structural parameters, specifically actuator height and surrounding
thickness, impact the bending angle and overall actuator performance. The results reveal
that increasing actuator height significantly enhances bending capability, while thicker
surrounding materials hinder bending, highlighting the need for careful design optimization.
Additionally, the study employs artificial neural networks to predict bending angles,
achieving an outstanding predictive accuracy with a residual variance of just 0.74% and an
explained variance of 99.26%. These results underscore the potential of machine learning
to refine actuator designs for therapeutic applications. The insights gained from this research
contribute to the development of improved design guidelines for soft actuators, advancing
rehabilitation technology and enabling more effective treatments for stroke patients.