Real-Time Bidirectional Haptic Data Forecasting in Teleoperated Telemedicine Systems

Teleoperation systems play a critical role in telemedicine by addressing the uneven distribution of medical resources and reducing infection risks. However, data transmission delays can cause tactile feedback lag, compromising the real-time performance of remote medical operations. To improve system position tracking accuracy and feedback fidelity, this study explores predictive techniques based on the Long-and Short-term Time-series Network (LSTNet), taking into account the differing characteristics of forward and backward data transmission. For forward transmission, a Soft-Modified Dynamic Time Warping (Soft-MDTW) pose prediction algorithm with a continuously differentiable search process is proposed, resulting in a 9.67% reduction in position tracking error. For backward transmission, a novel Haptic Structural Similarity Index Measure (HSSIM) loss function integrating tactile features is developed, improving the force signal similarity by 5.53%. These methods collectively enhance the real-time performance and safety of remote medical systems.