Evaluation of Musculoskeletal Aging and Related Disorders Via Advanced Clinical Imaging

Overview

Study Overview This clinical research focuses on the development and validation of a multimodal artificial intelligence (AI) platform designed for the automated diagnosis and precise staging of two major musculoskeletal conditions: Osteoporosis (OP) and Osteoarthritis (OA). By integrating diverse clinical imaging data, the study aims to provide a more objective and standardized approach to assessing bone and joint degeneration. Technological Core: Intelligent Staging Traditional diagnosis often relies on manual interpretation, which can lead to inter-observer variability. This study employs deep learning and multimodal imaging to: For Osteoporosis: Automatically quantify bone mineral density and micro-architectural changes to determine the stage of bone loss and evaluate fracture risk. For Osteoarthritis: Identify subtle radiological markers such as joint space narrowing and osteophyte formation to categorize the severity of joint degeneration according to international staging standards (e.g., Kellgren-Lawrence scale). Why This Matters Early Intervention: By identifying early-stage changes in bone density and joint integrity, clinicians can implement preventive treatments before significant disability occurs. Standardized Care: The intelligent diagnostic model provides a "digital second opinion," ensuring consistent staging across different healthcare settings. Efficiency: The automated workflow reduces the workload of radiologists while maintaining high diagnostic accuracy. Ethical Compliance The study is conducted at Peking University People's Hospital under the supervision of the Institutional Review Board (Approval No. 2026PHB097-001). It strictly adheres to international ethical standards, including the Declaration of Helsinki and Good Clinical Practice (GCP) guidelines, to ensure patient data privacy and safety.

1. Research Rationale and Goals Musculoskeletal aging often presents as a complex interplay between Osteoporosis (OP) and Osteoarthritis (OA). Despite their prevalence, current diagnostic workflows frequently treat these conditions in isolation, often relying on manual radiological staging that is prone to inter-observer variability. This study aims to develop and validate a multimodal artificial intelligence (AI) platform capable of simultaneous detection and precise disease staging for both OP and OA. By integrating diverse data sources-including clinical laboratory markers, patient history, and multiple imaging modalities (X-ray, CT, and MRI)-the project seeks to provide a holistic and objective assessment of skeletal health. 2. Study Design and Population The research employs a bidirectional observational cohort study design. Retrospective Cohort (Model Development): Data will be collected from approximately 1,500 patients who visited the Peking University People's Hospital (PKUPH) between November 2005 and November 2025. Prospective Cohort (External Validation): At least 500 new participants will be recruited starting from December 2025 to test the model's performance in a real-world clinical setting. The study targets adults aged 18 to 90 who have completed relevant musculoskeletal imaging scans. 3. Methodology and AI PipelineThe study is divided into three strategic phases:Phase I: Multimodal Data Integration: Collection of de-identified imaging (X-ray/CT/MR), laboratory indices (e.g., bone turnover markers, calcium-phosphorus metabolism), and clinical demographics (Age, BMI, medical history).Phase II: Intelligent Diagnostic Staging: Leveraging Convolutional Neural Networks (CNN) for image feature extraction and machine learning algorithms (e.g., XGBoost, SVM) for clinical feature fusion.For Osteoporosis: The model will categorize bone health stages (Normal, Osteopenia, Osteoporosis) using DXA T-scores as the gold standard.For Osteoarthritis: The system will automate grading based on the Kellgren-Lawrence (KL) scale, identifying joint space narrowing and osteophytic progression.Phase III: Validation and Explainability: Internal cross-validation and independent external testing using the prospective cohort. SHAP (Shapley Additive Explanations) analysis will be applied to quantify the contribution of each modality to the final diagnostic decision, ensuring clinical transparency. 4. Outcome Measures The primary outcome is the diagnostic accuracy (AUC, Sensitivity, Specificity) of the AI model for both conditions. Secondary outcomes focus on long-term clinical utility, including the incidence of new fragility fractures and changes in functional scores (e.g., VAS or OKS) during the follow-up periods (6, 12, 24, and 36 months). 5. Ethical Oversight and Data Safety The study is conducted at Peking University People's Hospital under the approval of the Institutional Review Board (Approval No. 2026PHB097-001). It adheres to GCP principles and the Declaration of Helsinki. All imaging and clinical data are strictly de-identified (anonymized) before being entered into the secure, encrypted research database to protect patient privacy.