Inteligencia artificial para el cribado de osteoporosis en radiografías panorámicas dentales: una revisión sistemática

Introduction: Osteoporosis is underdiagnosed because dual-energy X-ray absorptiometry, the reference standard, is underused, while dental panoramic radiographs are taken routinely and show mandibular cortical signs of low bone mineral density. We synthesized the diagnostic accuracy of artificial-intelligence models that screen for osteoporosis on these radiographs. Methods: We searched PubMed, OpenAlex, and Crossref for studies published from 2013 to 2025, screened 321 records after deduplication, retrieved full texts in open access, and included studies that evaluated an artificial-intelligence model against a bone-mineral-density reference or mandibular index. Accuracy was extracted per study and risk of bias appraised. Results: Ten studies met the eligibility criteria. Nine used dual-energy X-ray absorptiometry as the reference standard and eight employed deep convolutional networks. Reported area-under-the-curve values ranged from 0.76 to 0.89, and reported accuracy ranged from about 74% to 98%, with sensitivity mostly between 78% and 91% where a bone-density reference was used. Risk of bias was high or unclear in most studies, and only one study reported external validation. Conclusions: Artificial-intelligence models can detect osteoporosis on dental panoramic radiographs with promising accuracy, but the evidence is limited by small, single-centre datasets, heterogeneous methods, and a high risk of bias, so translation into opportunistic screening remains premature.