An abundance of medical data and enhanced computational power have led to a surge in Artificial Intelligence (AI) applications. Published studies involving AI in bone and osteoporosis research have increased exponentially, raising the need for transparent model development and reporting strategies. This review offers a comprehensive overview and systematic quality assessment of AI articles in osteoporosis while highlighting recent advancements. A systematic search in the PubMed database, from December 17th, 2020, to February 1st, 2023 was conducted to identify AI articles that relate to osteoporosis. The quality assessment of the studies relied on the systematic evaluation of 12 quality items derived from the MI-CLAIM checklist. The systematic search yielded 97 articles that fell into five areas; bone properties assessment (11 articles), osteoporosis classification (26 articles), fracture detection/classification (25 articles), risk prediction (24 articles) and bone segmentation (11 articles). The average quality score for each study area was 8.9 (range: 7-11) for bone properties assessment, 7.8 (range: 5-11) for osteoporosis classification, 8.4 (range: 7-11) for fracture detection, 7.6 (range: 4-11) for risk prediction, and 9.0 (range: 6-11) for bone segmentation. A 6th area, AI-driven clinical decision support, identified the studies from the five preceding areas which aimed to improve clinician efficiency, diagnostic accuracy and patient outcomes through AI-driven models and opportunistic screening by automating or assisting with specific clinical tasks in complex scenarios. The current work highlights disparities in study quality and a lack of standardized reporting practices. Despite these limitations, a wide range of models and examination strategies have shown promising outcomes to aid in the earlier diagnosis and improve clinical decision making. Through careful consideration of sources of bias in model performance assessment, the field can build confidence in AI-based approaches, ultimately leading to improved clinical workflows and patient outcomes.
This review covers the recent advancements in artificial intelligence (AI) for managing osteoporosis, an increasingly prevalent condition that weakens bone tissues and increases fracture risk. Analyzing 97 studies from December 2020 to February 2023, the present work highlights how AI enhances bone properties assessment, osteoporosis classification, fracture detection and classification, risk prediction, and bone segmentation. A systematic qualitative assessment of the studies revealed improvements in study quality compared with the earlier review period, supported by innovative and more explainable AI approaches. AI shows promise in clinical decision support by offering novel screening tools that can help in the earlier identification of the disease, improve clinical workflows and patient prognosis. New pre-processing strategies and advanced model architectures have played a critical role in these improvements. Researchers have enhanced the accuracy and predictive performance of traditional methods by integrating clinical data with imaging data through advanced multi-factorial AI techniques. These innovations, paired with standardized development and validation processes, promise to personalize medicine and enhance patient care in osteoporosis management.

BACKGROUND: With the progressive increase in aging populations, the use of opportunistic computed tomography (CT) scanning is increasing, which could be a valuable method for acquiring information on both muscles and bones of aging populations.
OBJECTIVE: The aim of this study was to develop and externally validate opportunistic CT-based fracture prediction models by using images of vertebral bones and paravertebral muscles.
METHODS: The models were developed based on a retrospective longitudinal cohort study of 1214 patients with abdominal CT images between 2010 and 2019. The models were externally validated in 495 patients. The primary outcome of this study was defined as the predictive accuracy for identifying vertebral fracture events within a 5-year follow-up. The image models were developed using an attention convolutional neural network-recurrent neural network model from images of the vertebral bone and paravertebral muscles.
RESULTS: The mean ages of the patients in the development and validation sets were 73 years and 68 years, and 69.1% (839/1214) and 78.8% (390/495) of them were females, respectively. The areas under the receiver operator curve (AUROCs) for predicting vertebral fractures were superior in images of the vertebral bone and paravertebral muscles than those in the bone-only images in the external validation cohort (0.827, 95% CI 0.821-0.833 vs 0.815, 95% CI 0.806-0.824, respectively; P<.001). The AUROCs of these image models were higher than those of the fracture risk assessment models (0.810 for major osteoporotic risk, 0.780 for hip fracture risk). For the clinical model using age, sex, BMI, use of steroids, smoking, possible secondary osteoporosis, type 2 diabetes mellitus, HIV, hepatitis C, and renal failure, the AUROC value in the external validation cohort was 0.749 (95% CI 0.736-0.762), which was lower than that of the image model using vertebral bones and muscles (P<.001).
CONCLUSIONS: The model using the images of the vertebral bone and paravertebral muscle showed better performance than that using the images of the bone-only or clinical variables. Opportunistic CT screening may contribute to identifying patients with a high fracture risk in the future.

Randomized trials have not been performed, and may never be, to determine if osteoporosis treatment prevents hip fracture in men. Addressing that evidence gap, we analyzed data from an observational study of new hip fractures in a large integrated healthcare system to compare the reduction in hip fractures associated with standard-of-care osteoporosis treatment in men versus women. Sampling from 271 389 patients age ≥ 65 who had a hip-containing computed tomography scan during care between 2005-2018, we selected all who subsequently had a first hip fracture (cases) after the CT scan (start of observation) and a sex-matched equal number of randomly selected patients. From those, we analyzed all who tested positive for osteoporosis (DXA-equivalent hip bone mineral density T-score ≤ -2.5, measured from the CT scan using VirtuOst). We defined \"treated\" as at least six months of any osteoporosis medication by prescription fill data during follow up; \"not-treated\" was no prescription fill. Sex-specific odds ratios of hip fracture for treated versus not-treated patients were calculated by logistic regression; adjustments included age, BMD T-score, a BMD-treatment interaction, body mass index, race/ethnicity, and seven baseline clinical risk factors. At two-year follow-up, 33.9% of the women (750/2211 patients) and 24.0% of the men (175/728 patients) were treated, primarily with alendronate; 51.3% and 66.3%, respectively, were not-treated; and 721 and 269, respectively, had a first hip fracture since the CT scan. Odds ratio of hip fracture for treated versus not-treated was 0.26 (95% confidence interval: 0.21-0.33) for women and 0.21 (0.13-0.34) for men; the ratio of these odds ratios (men:women) was 0.81 (0.47-1.37), indicating no significant sex effect. Various sensitivity and stratified analyses confirmed these trends, including results at five-year follow-up. Given these results and considering the relevant literature, we conclude that osteoporosis treatment prevents hip fracture similarly in both sexes.
Much evidence suggests that osteoporosis treatment should prevent hip fracture similarly in both sexes. However, because of their expense, randomized clinical trials to demonstrate that definitively have not been performed and may never be. As a result, osteoporosis testing and treatment is not as widely adopted for men as it is for women. Addressing that evidence gap, we analyzed data from over 250 000 patients in the Kaiser Permanente healthcare system in Southern California. Sampling a subset of all patients over a 13-year period who had had a computed tomography (CT or CAT) scan as part of their medical care for any reason, we measured bone mineral density from the CT scans to identify all patients who had osteoporosis at the hip and then used data from the electronic health records to determine statistically the risk of a future hip fracture for those who were treated for osteoporosis versus those who were not treated. We found that the reduction in risk of hip fracture associated with treatment did not differ between the sexes. These results demonstrate that treating osteoporosis in patients at high risk of hip fracture should reduce the risk of hip fracture similarly in both sexes.

OBJECTIVE: This study aimed to develop and validate a new model that focused on the risk of imminent vertebral fractures in women with osteoporosis.
METHODS: Data from 2,048 patients were extracted from three hospitals, of which 1,720 patients passed the inclusion and exclusion screen. The patients from Nanfang Hospital (NFH) were randomized at a 2:1 ratio to create a training cohort (n = 709) and an internal validation cohort (n = 355), with the patients from the other two hospitals (n = 656) used for external validation. The risk factors included in the imminent osteoporotic vertebral compression fractures (OVCFs) prediction model (labelled TVF) were sorted by the least absolute shrinkage and selection operator and constructed by logistic regression. The area under the receiver operating characteristic curve (AUC), the decision curve, and the clinical impact curves of the optimal model were analyzed to verify the model.
RESULTS: There were 138 and 161 fresh fractures in NFH and the other two hospitals, respectively. The lowest BMD T value and the history of vertebral fracture were integrated into the TVF model. The prediction power of TVF was demonstrated by the AUCs of 0.788 (95% confidence interval [CI], 0.728-0.849) in the training cohort and 0.774 (95% CI, 0.705-0.842) in the internal validation cohort, and 0.790 (95% CI, 0.742-0.839) and 0.741 (95% CI, 0.668-0.813) in the external validation cohorts.
CONCLUSIONS: The TVF model demonstrated good discrimination to stratify the imminent risk of OVCFs. We therefore consider the model as a pertinent commencement in the search for more accurate imminent OVCFs prediction.

An artificial intelligence-based case-finding strategy has been developed to systematically identify individuals with osteoporosis or at varying risk of fragility fracture. This strategy has the potential to close the critical care gap in osteoporosis treatment in primary care, thereby lessening the societal burden imposed by fragility fractures.
BACKGROUND: Osteoporotic fractures represent a major cause of morbidity and, in older adults, a precursor of disability, loss of independence, poor quality of life and premature death. Despite the detrimental health impact, osteoporosis remains largely underdiagnosed and undertreated worldwide. Subjects at risk for osteoporosis-related fractures are identified either via organised screening or case finding. In the absence of a population-based screening policy, subjects at high risk of fragility fractures are opportunistically identified when a fracture occurs or because of other clinical risk factors (CRFs) for osteoporotic fracture and areal bone mineral density (aBMD) measured by dual-energy X-ray absorptiometry (DXA).
OBJECTIVE: This paper describes the development of a novel case-finding strategy, named Osteoporosis Diagnostic and Therapeutic Pathway (ODTP), which enables to identify subjects with osteoporosis or at varying risk of fragility fracture. This strategy is based on a specifically designed software tool, named \"Bone Fragility Query\" (BFQ), which analyses the electronic health record (EHR) databases of General Practitioners (GPs) to systematically identify individuals who should be prescribed DXA-BMD measurement, vertebral fracture assessment (VFA) and anti-osteoporosis medications (AOM).
CONCLUSIONS: The ODTP through BFQ tool is a feasible, convenient and time-saving osteoporosis model of care for GPs during routine clinical practice. It enables GPs to shift their focus from what to do (clinical guidelines) to how to do it in the primary health care setting. It also allows a systematic approach to primary and secondary prevention of fragility fractures, thereby overcoming clinical inertia and contributing to closing the gap between evidence and practice for the management of osteoporosis in primary care.

UNASSIGNED: Osteoporotic fractures significantly impact individuals\'s quality of life and exert substantial pressure on the social pension system. This study aims to develop prediction models for osteoporotic fracture and uncover potential risk factors based on Electronic Health Records (EHR).
UNASSIGNED: Data of patients with osteoporosis were extracted from the EHR of Xinhua Hospital (July 2012-October 2017). Demographic and clinical features were used to develop prediction models based on 12 independent machine learning (ML) algorithms and 3 hybrid ML models. To facilitate a nuanced interpretation of the results, a comprehensive importance score was conceived, incorporating various perspectives to effectively discern and mine critical features from the data.
UNASSIGNED: A total of 8530 patients with osteoporosis were included for analysis, of which 1090 cases (12.8%) were fracture patients. The hybrid model that synergistically combines the Support Vector Machine (SVM) and XGBoost algorithms demonstrated the best predictive performance in terms of accuracy and precision (above 90%) among all benchmark models. Blood Calcium, Alkaline phosphatase (ALP), C-reactive Protein (CRP), Apolipoprotein A/B ratio and High-density lipoprotein cholesterol (HDL-C) were statistically found to be associated with osteoporotic fracture.
UNASSIGNED: The hybrid machine learning model can be a reliable tool for predicting the risk of fracture in patients with osteoporosis. It is expected to assist clinicians in identifying high-risk fracture patients and implementing early interventions.

Individuals with type 2 diabetes have lower trabecular bone score (TBS) and increased fracture risk despite higher bone mineral density. However, measures of trabecular microarchitecture from high-resolution peripheral computed tomography are not lower in type 2 diabetes. We hypothesized that confounding effects of abdominal tissue thickness may explain this discrepancy, since central obesity is a risk factor for diabetes and also artifactually lowers TBS. This hypothesis was tested in individuals aged 40 years and older from a large DXA registry, stratified by sex and diabetes status. When DXA-measured abdominal tissue thickness was not included as a covariate, men without diabetes had lower TBS than women without diabetes (mean difference -0.074, P < .001). TBS was lower in women with versus without diabetes (mean difference -0.037, P < .001), and men with versus without diabetes (mean difference -0.007, P = .042). When adjusted for tissue thickness these findings reversed, TBS became greater in men versus women without diabetes (mean difference +0.053, P < .001), in women with versus without diabetes (mean difference +0.008, P < .001), and in men with versus without diabetes (mean difference +0.014, P < .001). During mean 8.7 years observation, incident major osteoporotic fractures were seen in 7048 (9.6%). Adjusted for multiple covariates except tissue thickness, TBS predicted fracture in all subgroups with no significant diabetes interaction. When further adjusted for tissue thickness, HR per SD lower TBS remained significant and even increased slightly. In conclusion, TBS predicts fractures independent of other clinical risk factors in both women and men, with and without diabetes. Excess abdominal tissue thickness in men and individuals with type 2 diabetes may artifactually lower TBS using the current algorithm, which reverses after accounting for tissue thickness. This supports ongoing efforts to update the TBS algorithm to directly account for the effects of abdominal tissue thickness for improved fracture risk prediction.
Individuals with type 2 diabetes are at increased fracture risk despite having higher bone mineral density (BMD). Previous studies suggest that trabecular bone score (TBS), a measure of bone derived from spine DXA images that can be used to assess fracture risk in addition to BMD, may be lower in individuals with type 2 diabetes. However, TBS is artificially lowered by greater abdominal obesity. We showed that abdominal obesity explained the lower TBS measurements that were seen in individuals with type 2 diabetes. However, even when we considered the effect of abdominal obesity, TBS was still able to predict major fractures in both women and men, with and without diabetes.

Graphical Abstract.
Osteoporosis is an increasing burden for our aging society. Fracture risk assessment tool (FRAX) and areal bone mineral density (aBMD) have been mainly used as a surrogate, but only identify 46% of patients sustaining a hip fracture. Adding information about material and mechanical properties might improve the fracture risk prediction. In this study these properties were assessed of cortical and trabecular bone samples from the human femoral neck. In total, 178 trabeculae were obtained from 10 patients suffering a low-trauma fracture and 10 healthy donors (from a previous study) and 141 cortical specimens were newly manufactured from 17 low-trauma fracture patients and 15 controls. Cyclic tensile tests were performed to extract elastic, plastic, viscous, damage, and failure properties with a rheological model. No significant difference of any investigated property was determined. Interestingly, donor aBMD indicated a significant correlation with the post-yield behavior and damage accumulation (modulus degradation) of cortical bone. Cortical bone indicated a significantly larger apparent modulus (17.2 GPa), yield stress (50 MPa), viscosity (17.9 GPas), and damage accumulation (73%), but a decreased toughness (1.6 MJ/m3), than trabecular bone (8.8 GPa, 30 MPa, 9.3 GPas, 60%, 3.2 MJ/m3, respectively). Qualitatively, cortical bone displayed a linear-elastic phase, followed by a plastic phase with little post-yield hardening. In contrast, trabeculae yielded early, with a pronounced post-yield hardening phase and fractured at larger strains. Only a few correlations between donor mineral status and tissue mechanical behavior were found. It is suggested that the trabecularization of cortical bone with age and disease may not only result in a decreased bone mass, but further causes a transitioning from stiff elastic cortical to soft, viscous trabecular bone. This aspect warrants further investigation to determine its role in age- and osteoporosis-related bone fragility.

The novel metaPGS, integrating multiple fracture-related genetic traits, surpasses traditional polygenic scores in predicting fracture risk. Demonstrating a robust association with incident fractures, this metaPGS offers significant potential for enhancing clinical fracture risk assessment and tailoring prevention strategies.
BACKGROUND: Current polygenic scores (PGS) have limited predictive power for fracture risk. To improve genetic prediction, we developed and evaluated a novel metaPGS combining genetic information from multiple fracture-related traits.
METHODS: We derived individual PGS from genome-wide association studies of 16 fracture-related traits and employed an elastic-net logistic regression model to examine the association between the 16 PGSs and fractures. An optimal metaPGS was constructed by combining 11 significant individual PGSs selected by the elastic regularized regression model. We evaluated the predictive power of the metaPGS alone and in combination with clinical risk factors recommended by guidelines. The discrimination ability of metaPGS was assessed using the concordance index. Reclassification was assessed using net reclassification improvement (NRI) and integrated discrimination improvement (IDI).
RESULTS: The metaPGS had a significant association with incident fractures (HR 1.21, 95% CI 1.18-1.25 per standard deviation of metaPGS), which was stronger than previously developed bone mineral density (BMD)-related individual PGSs. Models with PGS_FNBMD, PGS_TBBMD, and metaPGS had slightly higher but statistically non-significant c-index than the base model (0.640, 0.644, 0.644 vs. 0.638). However, the reclassification analysis showed that compared to the base model, the model with metaPGS improves the reclassification of fracture.
CONCLUSIONS: The metaPGS is a promising approach for stratifying fracture risk in the European population, improving fracture risk prediction by combining genetic information from multiple fracture-related traits.

Whether simultaneous automated ascertainments of prevalent vertebral fracture (auto-PVFx) and abdominal aortic calcification (auto-AAC) on vertebral fracture assessment (VFA) lateral spine bone density (BMD) images jointly predict incident fractures in routine clinical practice is unclear. We estimated the independent associations of auto-PVFx and auto-AAC primarily with incident major osteoporotic and secondarily with incident hip and any clinical fractures in 11 013 individuals (mean [SD] age 75.8 [6.8] years, 93.3% female) who had a BMD test combined with VFA between March 2010 and December 2017. Auto-PVFx and auto-AAC were ascertained using convolutional neural networks (CNNs). Proportional hazards models were used to estimate the associations of auto-PVFx and auto-AAC with incident fractures over a mean (SD) follow-up of 3.7 (2.2) years, adjusted for each other and other risk factors. At baseline, 17% (n = 1881) had auto-PVFx and 27% (n = 2974) had a high level of auto-AAC (≥ 6 on scale of 0 to 24). Multivariable-adjusted hazard ratios (HR) for incident major osteoporotic fracture (95% CI) were 1.85 (1.59, 2.15) for those with compared with those without auto-PVFx, and 1.36 (1.14, 1.62) for those with high compared with low auto-AAC. The multivariable-adjusted HRs for incident hip fracture were 1.62 (95% CI, 1.26 to 2.07) for those with compared to those without auto-PVFx, and 1.55 (95% CI, 1.15 to 2.09) for those high auto-AAC compared with low auto-AAC. The 5-year cumulative incidence of major osteoporotic fracture was 7.1% in those with no auto-PVFx and low auto-AAC, 10.1% in those with no auto-PVFx and high auto-AAC, 13.4% in those with auto-PVFx and low auto-AAC, and 18.0% in those with auto-PVFx and high auto-AAC. While physician manual review of images in clinical practice will still be needed to confirm image quality and provide clinical context for interpretation, simultaneous automated ascertainment of auto-PVFx and auto-AAC can aid fracture risk assessment.
Individuals with calcification of their abdominal aorta (AAC) and vertebral fractures seen on lateral spine bone density images (easily obtained as part of a bone density test) are much more likely to have subsequent fractures. Prior studies have not shown if both AAC and prior vertebral fracture both contribute to fracture prediction in routine clinical practice. Additionally, a barrier to using these images to aid fracture risk assessment at the time of bone density testing has been the need for expert readers to be able to accurately detect both AAC and vertebral fractures. We have developed automated computer methods (using artificial intelligence) to accurately detect vertebral fracture (auto-PVFx) and auto-AAC on lateral spine bone density images for 11 013 older individuals having a bone density test in routine clinical practice. Over a 5-year follow-up period, 7.1% of those with no auto-PVFx and low auto-AAC, 10.1% of those with no auto-PVFx and high auto-AAC, 13.4% of those with auto-PVFx and low auto-AAC, and 18.0% of those with auto-PVFx and high auto-AAC had a major osteoporotic fracture. Auto-PVFx and auto-AAC, ascertained simultaneously on lateral spine bone density images, both contribute to the risk of subsequent major osteoporotic fractures in routine clinical practice settings.