Structure-aware fatigue modeling in foot deformities: A digital health framework for tissue-specific running injury risk prediction using multi-modal data.

Hallux Valgus (HV) is a common deformity in runners that reflects altered foot morphology and thereby redistributes mechanical loads along the lower limb, possibly increasing injury risk. However, the biomechanical consequences of this deformity on loading at common running injury locations, and the resulting fatigue-failure injury risk remain undocumented. This study integrated gait analysis with subject-specific musculoskeletal modelling to estimate Achilles tendon, plantar fascia, patellofemoral, and tibial loading and probability of fatigue-failure during running at 12 km/h in a cohort of 26 runners with HV and 26 healthy controls. HV runners exhibited pronounced structural deviation, reflected by increased hallux valgus and intermetatarsal angles, and elevated Foot Posture Index scores. During running, HV runners showed reduced ankle and metatarsophalangeal range of motions and increased peak joint moments (ankle p = 0.004; MTP p = 0.003). These biomechanical alterations were associated with higher Cumulative Load, Cumulative Damage, and Probability of Fatigue Failure, particularly in the Achilles tendon and plantar fascia, whereas tibial loading was largely similar to individuals without HV. To enable individualized prediction of biomechanical loading at the four injury locations, we trained deep learning models that used IMU input data (Model 1) or IMU plus foot structure data (Model 2). Model 1 (SSO-CNN-BiLSTM-HAM) accurately estimated loading indices (R² = 0.89-0.93), and Model 2 further enhanced predictive accuracy (R² = 0.94-0.97). Collectively, these findings suggest that HVA and foot posture index-related structural deviations increase lower-limb loading patterns and accelerate tissue fatigue-failure probability.