Joshua Robert

Background.  Fragility fractures are a massive public health crisis, causing roughly 2 million osteoporotic breaks every year in the U.S. alone.^[1] Currently, we rely on DXA scans and FRAX scores to find at-risk patients, but there’s a catch: about half of the people who actually suffer a fracture don’t meet the official criteria for osteoporosis.^[5]

The problem is that bone mineral density (BMD) measures how much bone you have, but not how good that bone is. It ignores crucial “quality” factors like pore size, cortical thickness, and how well the internal bone scaffolding is connected.^[1] While high-resolution CT (HR-pQCT) is the gold standard for looking at this architecture, it’s expensive and rarely available.^[2] This review explores “Digital Bone Age” (DBA), a new way to use AI to pull these hidden skeletal details from standard CT scans already being performed for other reasons.

Methods.  We conducted a structured search across PubMed and Google Scholar, focusing on the intersection of osteoporosis, deep learning, and opportunistic CT imaging. Search terms included “osteoporosis,” “fragility fracture,” “bone microarchitecture,” “opportunistic CT,” “deep learning,” “self-supervised learning,” and “fracture prediction.” We prioritized peer-reviewed research from 2006 to 2026 that looked at AI-driven fracture prediction and bone microarchitecture.

Results. The res earch is clear: bone structure predicts fractures better than density alone. Studies show that AI can analyze “opportunistic” CT scans (like a routine chest or abdominal CT) to measure vertebral strength and volume accurately.^[3,4]

Modern deep learning models are now outperforming traditional clinical risk factors by identifying patterns the human eye, and standard software, miss.^[6] Some of the most promising models even factor in surrounding muscle quality as a sign of overall frailty.^[8] Thanks to recent breakthroughs in “self-supervised learning,” AI can now standardize these measurements across different scanner brands, overcoming a massive technical hurdle.^[7] This suggests we can finally create a reliable, continuous “Digital Bone Age” metric from everyday imaging.^[9]

Conclusions. Our current tools are leaving too many vulnerable patients behind. By combining routine CT scans with ad vanced AI, we can see the “biological age” of a skeleton rather than just its mineral weight.

While the concept of DBA is promising, it still needs a formal definition and rigorous testing against DXA and FRAX. If future research can validate these models across different hospitals, DBA could turn every routine CT scan into a powerful, automated screening tool for fracture risk.^[9]

Works Cited

1. Seeman, E., & Delmas, P. D. (2006). Bone quality–the material and structural basis of bone strength and fragility. The New England journal of medicine, 354(21), 2250–2261. https://doi.org/10.1056/NEJMra053077

2. Boutroy, S., Bouxsein, M. L., Munoz, F., & Delmas, P. D. (2005). In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. The Journal of clinical endocrinology and metabolism, 90(12), 6508–6515. https://doi.org/10.1210/jc.2005-1258

3. Pickhardt, P. J., Lee, L. J., del Rio, A. M., Lauder, T., Bruce, R. J., Summers, R. M., Pooler, B. D., & Binkley, N. (2011). Simultaneous screening for osteoporosis at CT colonography: bone mineral density assessment using MDCT attenuation techniques compared with the DXA reference standard. Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research, 26(9), 2194–2203. https://doi.org/10.1002/jbmr.428

4. Adams, A. L., Fischer, H., Kopperdahl, D. L., Lee, D. C., Black, D. M., Bouxsein, M. L., Fatemi, S., Khosla, S., Orwoll, E. S., Siris, E. S., & Keaveny, T. M. (2018). Osteoporosis and Hip Fracture Risk From Routine Computed Tomography Scans: The Fracture, Osteoporosis, and CT Utilization Study (FOCUS). Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research, 33(7), 1291–1301. https://doi.org/10.1002/jbmr.3423

5. McCloskey, E. V., Odén, A., Harvey, N. C., Leslie, W. D., Hans, D., Johansson, H., Barkmann, R., Boutroy, S., Brown, J., Chapurlat, R., Elders, P. J. M., Fujita, Y., Glüer, C. C., Goltzman, D., Iki, M., Karlsson, M., Kindmark, A., Kotowicz, M., Kurumatani, N., Kwok, T., … Kanis, J. A. (2016). A Meta-Analysis of Trabecular Bone Score in Fracture Risk Prediction and Its Relationship to FRAX. Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research, 31(5), 940–948. https://doi.org/10.1002/jbmr.2734

6. Kang, J. W., Park, C., Lee, D. E., Yoo, J. H., & Kim, M. (2023). Prediction of bone mineral density in CT using deep learning with explainability. Frontiers in physiology, 13, 1061911. https://doi.org/10.3389/fphys.2022.1061911

7. Guha, I., Nadeem, S. A., Zhang, X., DiCamillo, P. A., Levy, S. M., Wang, G., & Saha, P. K. (2024). Deep learning-based harmonization of trabecular bone microstructures between high- and low-resolution CT imaging. Medical physics, 51(6), 4258–4270. https://doi.org/10.1002/mp.17003

8. Oh, S., Kang, W. Y., Park, H., Yang, Z., Lee, J., Kim, C., Woo, O. H., & Hong, S. J. (2024). Evaluation of deep learning-based quantitative computed tomography for opportunistic osteoporosis screening. Scientific reports, 14(1), 363. https://doi.org/10.1038/s41598-023-45824-7

9. Cheng, H., Zhang, Y., Meng, M., Liu, S., Yang, Y., Ran, Y., & Kou, Y. (2026). Artificial intelligence in osteoporosis assessment using CT imaging: a scoping review. Frontiers in medicine, 13, 1779483. https://doi.org/10.3389/fmed.2026.1779483