SCH: Digital Twin Systems for Personalized Prediction of Osteoporotic Spine - PROJECT SUMMARY (See instructions): Osteoporosis weakens vertebral bone by reducing bone mass, deteriorating trabecular microstructure, and altering spatial stiffness distributions, increasing the risk of vertebral fractures that cause pain, deformity, disability, loss of independence, and mortality in older adults. Current clinical tools estimate population-level fracture probability, but they do not predict how patient-specific vertebral geometry, trabecular deterioration, heterogeneous stiffness, and daily activity loading combine to create localized strains that precede fracture. The long-term objective is to advance predictive and mechanistic understanding of osteoporotic vertebral fracture and translate this knowledge into a patient-specific, physics-informed digital twin framework for fracture prediction, prevention, and spine treatment planning. The central hypothesis is that CT-derived bone microstructure and patient covariates, constrained by elasticity and fracture physics and grounded in cadaveric deformation data, can support reliable estimation of heterogeneous vertebral stiffness and accurate prediction of localized, activity-dependent fracture risk with quantified uncertainty. The specific aims are to (1) infer three-dimensional, spatially varying vertebral elasticity and fracture-related descriptors from cadaveric CT images and digital volume correlation displacement fields; (2) develop personalized Bayesian inference methods that link CT-derived bone microarchitecture and clinical covariates to patient-specific vertebral material properties with uncertainty; (3) build a scalable vertebral digital twin that predicts localized strain and fracture-risk patterns from patient-specific bone-property maps; and (4) extend the framework to whole-spine, activity-specific prediction by linking motion-derived vertebral loading during daily activities to vertebra-level strain and fracture probability. The research design integrates cadaveric vertebral testing, CT imaging, digital volume correlation, finite element modeling, musculoskeletal load estimation, Bayesian physics-informed machine learning, and blinded clinical review. Expected outcomes include validated methods for estimating patient-specific vertebral material properties, uncertainty-aware fracture-risk maps, and activity-specific risk predictions. By using these predictions to identify why, where, and under what loading conditions osteoporotic vertebrae become mechanically vulnerable , this project will generate new biomechanical knowledge of vertebral fragility and directly advance the NIAMS mission.
