A comprehensive evaluation of seven osteoporosis screening tools.

UNLABELLED: This review evaluates the sensitivity, specificity, and predictive values of seven osteoporosis screening tools in populations such as postmenopausal women, older adults, and patients with specific diseases. These tools offer advantages, including high sensitivity (facilitating early detection of osteoporosis), high cost-effectiveness, and the ability to be customized according to disease characteristics. However, they commonly suffer from low specificity, leading to a large number of subjects being misclassified as high-risk. This review identifies that the application of machine learning holds promise for improving the performance of screening tools. By integrating multidimensional data and processing large-scale sample data, machine learning is expected to overcome traditional limitations and comprehensively improve the efficiency of osteoporosis screening. In the future, integrating multimodal artificial intelligence with electronic health records may enable more accurate, comprehensive prediction of osteoporosis. This review summarizes the current strengths and limitations of osteoporosis screening tools and proposes directions for optimization and future research. OBJECTIVE: This study focuses on seven osteoporosis screening tools: Osteoporosis Self-Assessment Tool for Asians (OSTA), Osteoporosis Self-Assessment Tool (OST), Osteoporosis Risk Assessment Instrument (ORAI), Simple Calculated Osteoporosis Risk Estimation (SCORE), A Simple Bone Health Evaluation Tool (ABONE), Simple Osteoporosis Fracture Risk Tool for Seniors (SOFSURF), and Osteoporosis Self-Report Instrument for Identification of Seniors at Risk (OSIRIS). It summarizes key indicators such as their sensitivity, specificity, positive predictive value, and negative predictive value, aiming to evaluate their application performance in different populations, including postmenopausal women, older adults, and patients with specific diseases, and discusses the current advantages and dilemmas of osteoporosis screening tools. METHOD: A literature search was conducted in PubMed, Web of Science, and CNKI for studies on the seven target osteoporosis screening tools published over the past 25 years. The inclusion criteria included peer-reviewed original research articles (cross-sectional studies, prospective cohort studies, and diagnostic validation studies), excluding reviews, meta-analyses, animal studies, non-clinical studies, and non-English articles. RESULT: Most of the seven screening tools demonstrate high sensitivity, facilitating early detection of osteoporosis and offering advantages such as cost-effectiveness and adaptability to diverse disease characteristics. However, these tools have obvious limitations, primarily low specificity, which often leads to false-positive results. Machine learning shows potential to improve screening performance by integrating multidimensional data and processing large-scale datasets, though the evidence remains preliminary and requires further validation. The integration of multimodal artificial intelligence and electronic health records is proposed as a future direction, but it requires further validation. CONCLUSION: Based on existing evidence, the advantages and limitations of seven osteoporosis screening tools were summarized. It was noted that more high-quality research is needed to validate novel screening strategies (such as machine learning-based models) before their widespread clinical application.