Spinal metastatic bone disease (SMBD) affects approximately 30–70% of cancer patients worldwide. By compromising vertebral structural integrity, SMBD markedly increases fracture risk, potentially leading to spinal cord injury. In severe cases, management requires surgical intervention followed by a prolonged recovery of 6-8 months, an onerous burden for patients with limited life expectancy. Despite this, clinically actionable tools for patient-specific fracture risk assessment remain scarce. To address this gap, we propose a computational framework that uses patient-specific simulations to quantify load distribution and consequently fracture risk in vertebrae with progressive metastatic involvement. From routine CT imaging, our pipeline generates patient-specific vertebral geometries through a deep-learning workflow: (1) a U-Net localises the spinal column, (2) a Spatial Configuration Network identifies vertebral centroids, and (3) a Statistical Shape Model (SSM) performs individual vertebra segmentation. The SSM is trained on hundreds of CT scans, excluding metastatic cases. Each resulting segmentation is a random variable within the learned SSM distribution, enabling quantitative quality control via the Mahalanobis distance. Segmentations exceeding a predefined threshold are flagged as erroneous, while valid cases proceed through the pipeline. Accepted segmentations are tetrahedralized to produce homogeneous volumetric meshes, with relevant anatomical surfaces labelled. The meshes are then imported into Firedrake, a Python-based finite element solver, and physiological loading is simulated, modelling bone and tumour tissue as distinct homogeneous, isotropic, elastic–plastic materials. As a proof-of-concept, we apply the framework to the lumbar spine. This fully automated pipeline is designed for deployment across large CT databases, providing quantitative confidence measures at each processing stage. The long-term objective is to extend the framework to the entire spine and validate it using SMBD datasets. Ultimately, this work aims to deliver a robust, quantitative tool to support clinical decision-making in vertebral fracture risk assessment.