Osteoporosis is a systemic bone disease characterized by decreased bone strength, making people more susceptible to fractures, especially in the elderly. This project aims to enhance opportunistic screening for osteoporosis by integrating machine learning techniques and causal reasoning methods to identify patients at risk of osteoporosis early, so that clinical decisions can be made earlier and the incidence of serious illness can be reduced. Advanced machine learning models will be used to extract relevant features from medical images to accurately predict bone mineral density (BMD) and identify individuals at higher risk of osteoporosis. In addition, a causal relationship framework will be used to reveal the causal relationship between image-derived features and detect potential confounders that may affect the causal pathway to ensure unbiasedness. By generating counterfactual scenarios, the project aims to predict the impact of various interventions on the progression of osteoporosis in patients. The overall goal of this project is to change the clinical practice of osteoporosis management, promote timely intervention, and reduce the burden of osteoporosis-related fractures.
