Disentangling history and propagation dependencies in cross-subject knee contact stress prediction using a shared MeshGraphNet backbone.

Background.Subject-specific finite element analysis accurately characterizes knee joint mechanics but is computationally expensive. Deep surrogate models provide a near-real-time alternative, yet their generalization across subjects under limited pose and load inputs remains unclear. It remains unclear whether prediction uncertainty under limited pose/load inputs is more strongly associated with temporal history dependence or with spatial propagation dependence.Methods.To disentangle these factors, we employed a shared MeshGraphNet (MGN) backbone under a shared reference anatomy and mesh topology. A dataset of running trials from nine subjects was constructed using an OpenSim-FEBio workflow. We developed four model variants to isolate specific dependencies: (1) a baseline MGN; (2) CT-MGN, incorporating a control Transformer (CT) to encode short-horizon history; (3) MsgModMGN, applying state-conditioned modulation to message passing for adaptive propagation; and (4) CT-MsgModMGN, combining both mechanisms. Models were evaluated using a rigorous grouped 3-fold cross-validation on unseen subjects.Results.The models incorporating history encoding (CT-MGN and CT-MsgModMGN) significantly outperformed the baseline MGN and MsgModMGN in global accuracy (lower root-mean-square error) and spatial consistency (higher Dice). Crucially, the CT-based models mitigated the peak-shaving tendency commonly observed in deep surrogates and reduced peak-stress reconstruction error (REmax). In contrast, the spatial propagation modulation message modulation (MsgMod) alone yielded no significant improvement over the baseline, and combining it with CT provided no additional benefit.Conclusion.Temporal-history-related limitations contributed more strongly to prediction uncertainty than spatial propagation modulation in the present setting. Explicitly encoding short-horizon driver sequences enables the surrogate model to capture useful temporal context related to phase/state transitions, thereby achieving superior fidelity in peak-stress capture and high-risk-region localization compared to purely state-based approaches.