OPPORTUNISTIC OSTEOPOROSIS SCREENING USING ARTIFICIAL INTELLIGENCE IN ROUTINE COMPUTED TOMOGRAPHY: AN INTEGRATIVE LITERATURE REVIEW

Abstract

Osteoporosis is a common metabolic bone disease characterized by reduced bone strength and an increased risk of fragility fractures. Despite its clinical and social burden, it remains widely underdiagnosed, particularly before the first fracture occurs. This integrative review critically analyzed scientific evidence on the application of artificial intelligence to opportunistic screening for osteoporosis, low bone mineral density, and vertebral fractures using routine computed tomography. PubMed/MEDLINE, SciELO, and LILACS/BVS were searched for original studies published from January 2021 to July 2026 in Portuguese, English, or Spanish, involving adults and artificial intelligence, machine learning, deep learning, radiomics, or automated algorithms applied to body CT. A total of 1,432 records were identified, and 18 original studies were included in the final synthesis. The models showed promising performance for automated bone mineral density estimation, osteoporosis and osteopenia classification, vertebral fracture detection, and future fracture risk prediction. However, retrospective designs predominated, with methodological heterogeneity and limited external validation. Artificial intelligence shows substantial potential as a support tool for opportunistic screening, although widespread adoption requires standardization, multicenter validation, and evidence of clinical impact.