<p>Modeling the nonlinear dynamics of deformable linear objects under complex contact conditions remains a challenging problem in multibody system simulation. Traditional geometric approaches often neglect internal elastic effects, while standard Cosserat rod models with fixed contact parameterization may show limited accuracy when local frictional interactions and contact modes evolve rapidly. This paper presents a tactile-informed contact dynamics model that integrates local tactile feedback into a discrete Cosserat rod formulation for online state prediction in robotic cable manipulation. The proposed method introduces tactile interaction through equivalent contact quantities and a mode dependent state correction step within the dynamic update, thereby improving prediction consistency during contact-rich maneuvers. The model is evaluated against purely geometric and standard dynamic baselines across three representative scenarios: lift-off, sliding, and high curvature bending. Numerical results show that the proposed model reduces position and orientation errors and improves contact region consistency, with the most evident gains observed in the lift-off and high curvature bending scenarios. In addition, although the model does not satisfy the strict real time performance ratio condition at millisecond-scale integration steps, its single-step computation time (2.8 ms) remains well within the latency budget of a typical robotic control cycle (50&#xa0;ms). By incorporating fingertip tactile measurements into Cosserat rod dynamics, the proposed method provides a computationally feasible approach to contact-aware online cable state prediction.</p>