Exploratory bone-brain transcriptomic associations and an ODE-based digital-twin framework for fracture healing

Fracture healing is generally viewed as a local tissue-regeneration process, although skeletal injury can also elicit systemic immune, vascular, and neural responses. We used a stratified analytical strategy combining reanalysis of public transcriptomic datasets, descriptive single-nucleus profiling, and an ordinary differential equation (ODE)-based simulation of fracture healing to examine possible associations between adverse bone phenotypes and Alzheimer disease. In the primary random-effects meta-analysis of three retained human bone-related binary cohorts, 72 prespecified panel genes were measured in all three cohorts; BMP10 and ANGPT1 had the smallest adjusted P values. In the separate GSE48350 whole-brain dataset, 6 of 84 mapped target genes had nominal P < 0.05. Descriptive single-nucleus RNA-sequencing analysis of GSE138852 showed within-dataset, cell-type-specific expression patterns across annotated neural cell populations. The eight-state ODE prototype, built on prespecified assumptions, generated fracture-healing trajectories and scenario-dependent model outputs. Tissue-expression and genetic analyses provided additional exploratory context for the selected genes. These analyses define a hypothesis-generating framework for studying skeletal-neural associations after fracture.