Search

UNLABELLED: Hip fracture risk assessment is an important but challenging task. Quantitative CT-based patient-specific finite element (FE) analysis (FEA) incorporates bone geometry and bone density in the proximal femur. We developed a global FEA-computed fracture risk index to increase the prediction accuracy of hip fracture incidence. PURPOSE: Quantitative CT-based patient-specific finite element (FE) analysis (FEA) incorporates bone geometry and bone density in the proximal femur to compute the force (fracture load) and energy necessary to break the proximal femur in a particular loading condition. The fracture loads and energies-to-failure are individually associated with incident hip fracture, and provide different structural information about the proximal femur. METHODS: We used principal component analysis (PCA) to develop a global FEA-computed fracture risk index that incorporates the FEA-computed yield and ultimate failure loads and energies-to-failure in four loading conditions of 110 hip fracture subjects and 235 age- and sex-matched control subjects from the AGES-Reykjavik study. Using a logistic regression model, we compared the prediction performance for hip fracture based on the stratified resampling. RESULTS: We referred the first principal component (PC1) of the FE parameters as the global FEA-computed fracture risk index, which was the significant predictor of hip fracture (p-value

We aim to develop a deep-learning-based method for automatic proximal femur segmentation in quantitative computed tomography (QCT) images. We proposed a spatial transformation V-Net (ST-V-Net), which contains a V-Net and a spatial transform network (STN) to extract the proximal femur from QCT images. The STN incorporates a shape prior into the segmentation network as a constraint and guidance for model training, which improves model performance and accelerates model convergence. Meanwhile, a multi-stage training strategy is adopted to fine-tune the weights of the ST-V-Net. We performed experiments using a QCT dataset which included 397 QCT subjects. During the experiments for the entire cohort and then for male and female subjects separately, 90% of the subjects were used in ten-fold stratified cross-validation for training and the rest of the subjects were used to evaluate the performance of models. In the entire cohort, the proposed model achieved a Dice similarity coefficient (DSC) of 0.9888, a sensitivity of 0.9966 and a specificity of 0.9988. Compared with V-Net, the Hausdorff distance was reduced from 9.144 to 5.917 mm, and the average surface distance was reduced from 0.012 to 0.009 mm using the proposed ST-V-Net. Quantitative evaluation demonstrated excellent performance of the proposed ST-V-Net for automatic proximal femur segmentation in QCT images. In addition, the proposed ST-V-Net sheds light on incorporating shape prior to segmentation to further improve the model performance.

The purpose of this study was to determine the influence of frontal and transverse plane rotations of the femur and tibia on peak maximum principal stress in the patellar tendon. Using finite element modeling, patellar tendon stress profiles of eight healthy individuals were obtained during a simulated squatting task (45° of knee flexion). The femur and tibia of each model were rotated 10° (in 2° increments) along their respective axes beyond that of the natural degree of rotation. This process was repeated for the transverse plane (internal and external rotation) and frontal plane (adduction and abduction). Quasi-static loading simulations were performed to quantify peak maximum principal stress in patellar tendon. Internal and external rotations of the femur and tibia that exceeded 4° beyond that of the natural rotation resulted in progressively greater patellar tendon stress (p

BackgroundHip fracture occurs when an applied force exceeds the force that the proximal femur can support (the fracture load or "strength") and can have devastating consequences with poor functional outcomes. Proximal femoral strengths for specific loading conditions can be computed by subject-specific finite element analysis (FEA) using quantitative computerized tomography (QCT) images. However, the radiation and availability of QCT limit its clinical usability. Alternative low-dose and widely available measurements, such as dual energy X-ray absorptiometry (DXA) and genetic factors, would be preferable for bone strength assessment. The aim of this paper is to design a deep learning-based model to predict proximal femoral strength using multi-view information fusion.ResultsWe developed new models using multi-view variational autoencoder (MVAE) for feature representation learning and a product of expert (PoE) model for multi-view information fusion. We applied the proposed models to an in-house Louisiana Osteoporosis Study (LOS) cohort with 931 male subjects, including 345 African Americans and 586 Caucasians. We performed genome-wide association studies (GWAS) to select 256 genetic variants with the lowest p-values for each proximal femoral strength and integrated whole genome sequence (WGS) features and DXA-derived imaging features to predict proximal femoral strength. The best prediction model for fall fracture load was acquired by integrating WGS features and DXA-derived imaging features. The designed models achieved the mean absolute percentage error of 18.04%, 6.84% and 7.95% for predicting proximal femoral fracture loads using linear models of fall loading, nonlinear models of fall loading, and nonlinear models of stance loading, respectively.ConclusionThe proposed models are capable of predicting proximal femoral strength using WGS features and DXA-derived imaging features. Though this tool is not a substitute for predicting FEA using QCT images, it would make improved assessment of hip fracture risk more widely available while avoiding the increased radiation exposure from QCT.

BackgroundOsteoporosis is a progressive bone disease that is characterized by a decrease in bone mass and the deterioration in bone microarchitecture, which might be related to age and space travel. An unmet need exists for the development of novel imaging technologies to characterize osteoporosis.PurposeThe purpose of our study is to investigate the feasibility of X-ray-induced acoustic computed tomography (XACT) imaging for osteoporosis detection.MethodsAn in-house simulation workflow was developed to assess the ability of XACT for osteoporosis detection. To evaluate this simulation workflow, a three-dimensional digital bone phantom for XACT imaging was created by a series of two-dimensional micro-computed tomography (micro-CT) slices of normal and osteoporotic bones in mice. In XACT imaging, the initial acoustic pressure rise caused by the X-ray induce acoustic (XA) effect is proportional to bone density. First, region growing was deployed for image segmentation of different materials inside the bone. Then k-wave simulations were deployed to model XA wave propagation, attenuation, and detection. Finally, the time-varying pressure signals detected at each transducer location were used to reconstruct the XACT image with a time-reversal reconstruction algorithm.ResultsThrough the simulated XACT images, cortical porosity has been calculated, and XA signal spectra slopes have been analyzed for the detection of osteoporosis. The results have demonstrated that osteoporotic bones have lower bone mineral density and higher spectra slopes. These findings from XACT images were in good agreement with porosity calculation from micro-CT images.ConclusionThis work explores the feasibility of using XACT imaging as a new imaging tool for Osteoporosis detection. Considering that acoustic signals are generated by X-ray absorption, XACT imaging can be combined with traditional X-ray imaging that holds potential for clinical management of osteoporosis and other bone diseases.

Comprehensive treatment for vertebral metastatic lesions commonly involves vertebral augmentation (vertebroplasty or kyphoplasty) to relieve pain and stabilize the spine followed by multiple sessions of radiotherapy. We propose to combine vertebral augmentation and radiotherapy into a single treatment by adding32P, aβ-emitting radionuclide, to bone cement, thereby enabling spinal brachytherapy to be performed without irradiating the spinal cord. The goal of this study was to address key dosimetry and safety questions prior to performing extensive animal studies. The32P was in the form of hydroxyapatite powder activated by neutron bombardment in a nuclear reactor. We performedex vivodosimetry experiments to establish criteria for safe placement of the cement within the sheep vertebral body. In anin vivostudy, we treated three control ewes and three experimental ewes with brachytherapy cement containing 2.23-3.03 mCi32P ml-1to identify the preferred surgical approach, to determine if32P leaches from the cement and into the blood, urine, or feces, and to identify unexpected adverse effects. Ourex vivoexperiments showed that cement with 4 mCi32P ml-1could be safely implanted in the vertebral body if the cement surface is at least 4 mm from the spinal cord in sheep and 5 mm from the spinal cord in humans.In vivo, a lateral retroperitoneal surgical approach, ventral to the transverse processes, was identified as easy to perform while allowing a safe distance to the spinal cord. The blood, urine, and feces of the sheep did not contain detectable levels of32P, and the sheep did not experience any neurologic or other adverse effects from the brachytherapy cement. These results demonstrate, on a preliminary level, the relative safety of this brachytherapy cement and support additional development and testing.

Physical activity enhances proximal femur bone mass, but it remains unclear whether the benefits translate into an enhanced ability to resist fracture related loading. We recently used baseball pitchers as a within-subject controlled model to demonstrate physical activity induced proximal femur adaptation in regions associated with weight bearing and increased strength under single-leg stance loading. However, there was no measurable benefit to resisting common injurious loading (e.g. a fall onto the greater trochanter). A lack of power and a small physical activity effect size may have contributed to the latter null finding. Softball pitchers represent an alternative within-subject controlled model to explore adaptation of the proximal femur to physical activity, exhibiting greater dominant-to-nondominant (D-to-ND) leg differences than baseball pitchers. The current study used quantitative computed tomography, statistical parametric mapping, and subject-specific finite element (FE) modeling to explore adaptation of the proximal femur to physical activity in female softball pitchers (n = 25). Female cross-country runners (n = 15) were included as symmetrically loaded controls, showing very limited D-to-ND leg differences. Softball pitchers had D-to-ND leg differences in proximal femur, femoral neck, and trochanteric volumetric bone mineral density and content, and femoral neck volume. Voxel-based morphometry analyses and cortical bone mapping showed D-to-ND leg differences within a large region connecting the superior femoral head, inferior femoral neck and medial intertrochanteric region, and within the greater trochanter. FE modeling revealed pitchers had 19.4% (95%CI, 15.0 to 23.9%) and 4.9% (95%CI, 1.7 to 8.2%) D-to-ND leg differences in predicted ultimate strength under single-leg stance loading and a fall onto the greater trochanter, respectively. These data affirm the spatial and strength adaptation of the proximal femur to weight bearing directed loading and demonstrate that the changes can also have benefits, albeit smaller, on resisting loads associated with a sideways fall onto the greater trochanter.

Physical activity (PA) enhances proximal femur bone mass, as assessed using projectional imaging techniques. However, these techniques average data over large volumes, obscuring spatially heterogeneous adaptations. The current study used quantitative computed tomography, statistical parameter mapping, and subject-specific finite element (FE) modeling to explore spatial adaptation of the proximal femur to PA. In particular, we were interested in adaptation occurring at the superior femoral neck and improving strength under loading from a fall onto the greater trochanter. High/long jump athletes (n = 16) and baseball pitchers (n = 16) were utilized as within-subject controlled models as they preferentially load their take-off leg and leg contralateral to their throwing arm, respectively. Controls (n = 15) were included but did not show any dominant-to-nondominant (D-to-ND) leg differences. Jumping athletes showed some D-to-ND leg differences but less than pitchers. Pitchers had 5.8% (95% confidence interval [CI] 3.9%-7.6%) D-to-ND leg differences in total hip volumetric bone mineral density (vBMD), with increased vBMD in the cortical compartment of the femoral neck and trochanteric cortical and trabecular compartments. Voxel-based morphometry analyses and cortical bone mapping showed pitchers had D-to-ND leg differences within the regions of the primary compressive trabeculae, inferior femoral neck, and greater trochanter but not the superior femoral neck. FE modeling revealed pitchers had 4.1% (95% CI 1.4%-6.7%) D-to-ND leg differences in ultimate strength under single-leg stance loading but no differences in ultimate strength to a fall onto the greater trochanter. These data indicate the asymmetrical loading associated with baseball pitching induces proximal femur adaptation in regions associated with weight bearing and muscle contractile forces and increases strength under single-leg stance loading. However, there were no benefits evident at the superior femoral neck and no measurable improvement in ultimate strength to common injurious loading during aging (ie, fall onto the greater trochanter), raising questions as to how to better target these variables with PA. © 2019 American Society for Bone and Mineral Research.

Concerns raised at a 2010 Bone Summit held for National Aeronautics and Space Administration Johnson Space Center led experts in finite element (FE) modeling for hip fracture prediction to propose including hip load capacity in the standards for astronaut skeletal health. The current standards for bone are based upon areal bone mineral density (aBMD) measurements by dual X-ray absorptiometry (DXA) and an adaptation of aBMD cut-points for fragility fractures. Task Group members recommended (i) a minimum permissible outcome limit (POL) for post-mission hip bone load capacity, (ii) use of FE hip load capacity to further screen applicants to astronaut corps, (iii) a minimum pre-flight standard for a second long-duration mission, and (iv) a method for assessing which post-mission physical activities might increase an astronaut's risk for fracture after return. QCT-FE models of eight astronaut were analyzed using nonlinear single-limb stance (NLS) and posterolateral fall (NLF) loading configurations. QCT data from the Age Gene/Environment Susceptibility (AGES) Reykjavik cohort and the Rochester Epidemiology Project were analyzed using identical modeling procedures. The 75th percentile of NLS hip load capacity for fractured elderly males of the AGES cohort (9537N) was selected as a post-mission POL. The NLF model, in combination with a Probabilistic Risk Assessment tool, was used to assess the likelihood of exceeding the hip load capacity during post-flight activities. There was no recommendation to replace the current DXA-based standards. However, FE estimation of hip load capacity appeared more meaningful for younger, physically active astronauts and was recommended to supplement aBMD cut-points.

Purpose of reviewPhysical activity improves proximal femoral bone health; however, it remains unclear whether changes translate into a reduction in fracture risk. To enhance any fracture-protective effects of physical activity, fracture prone regions within the proximal femur need to be targeted.Recent findingsThe proximal femur is designed to withstand forces in the weight-bearing direction, but less so forces associated with falls in a sideways direction. Sideways falls heighten femoral neck fracture risk by loading the relatively weak superolateral region of femoral neck. Recent studies exploring regional adaptation of the femoral neck to physical activity have identified heterogeneous adaptation, with adaptation principally occurring within inferomedial weight-bearing regions and little to no adaptation occurring in the superolateral femoral neck. There is a need to develop novel physical activities that better target and strengthen the superolateral femoral neck within the proximal femur. Design of these activities may be guided by subject-specific musculoskeletal modeling and finite-element modeling approaches.

BackgroundQuantitative computed tomography (QCT) imaging is the basis for multiple assessments of bone quality in the proximal femur, including volumetric bone mineral density (vBMD), tissue volume, estimation of bone strength using finite element modeling (FEM), cortical bone thickness, and computational-anatomy-based morphometry assessments.MethodsHere, we present an automatic framework to perform a multi-parametric QCT quantification of the proximal femur. In this framework, the proximal femur is cropped from the bilateral hip scans, segmented using a multi-atlas based segmentation approach, and then assigned volumes of interest through the registration of a proximal femoral template. The proximal femur is then subjected to compartmental vBMD, compartmental tissue volume, FEM bone strength, compartmental surface-based cortical bone thickness, compartmental surface-based vBMD, local surface-based cortical bone thickness, and local surface-based cortical vBMD computations. Consequently, the template registrations together with vBMD and surface-based cortical bone parametric maps enable computational anatomy studies. The accuracy of the segmentation was validated against manual segmentations of 80 scans from two clinical facilities, while the multi-parametric reproducibility was evaluated using repeat scans with repositioning from 22 subjects obtained on CT imaging systems from two manufacturers.ResultsAccuracy results yielded a mean dice similarity coefficient of 0.976±0.006, and a modified Haussdorf distance of 0.219±0.071 mm. Reproducibility of QCT-derived parameters yielded root mean square coefficients of variation (CVRMS) between 0.89-1.66% for compartmental vBMD; 0.20-1.82% for compartmental tissue volume; 3.51-3.59% for FEM bone strength; 1.89-2.69% for compartmental surface-based cortical bone thickness; and 1.08-2.19% for compartmental surface-based cortical vBMD. For local surface-based assessments, mean CVRMS were between 3.45-3.91% and 2.74-3.15% for cortical bone thickness and vBMD, respectively.ConclusionsThe automatic framework presented here enables accurate and reproducible QCT multi-parametric analyses of the proximal femur. Our subjects were elderly, with scans obtained across multiple clinical sites and manufacturers, thus documenting its value for clinical trials and other multi-site studies.

In multicenter studies and longitudinal studies that use two or more different quantitative computed tomography (QCT) imaging systems, anthropomorphic standardization phantoms (ASPs) are used to correct inter-scanner differences and allow pooling of data. In this study, in vivo imaging of 20 women on two imaging systems was used to evaluate inter-scanner differences in hip integral BMD (iBMD), trabecular BMD (tBMD), cortical BMD (cBMD), femoral neck yield moment (My) and yield force (Fy), and finite-element derived strength of the femur under stance (FEstance) and fall (FEfall) loading. Six different ASPs were used to derive inter-scanner correction equations. Significant (p

To improve bone strength prediction beyond limitations of assessment founded solely on the bone mineral component, we investigated the effect of hyperlipidemia, present in more than 40% of osteoporotic patients, on multiscale structure of murine bone. Our overarching purpose is to estimate bone strength accurately, to facilitate mitigating fracture morbidity and mortality in patients. Because (i) orientation of collagen type I affects, independently of degree of mineralization, cortical bone׳s micro-structural strength; and, (ii) hyperlipidemia affects collagen orientation and μCT volumetric tissue mineral density (vTMD) in murine cortical bone, we have constructed the first multiscale finite element (mFE), mouse-specific femoral model to study the effect of collagen orientation and vTMD on strength in Ldlr(-/-), a mouse model of hyperlipidemia, and its control wild type, on either high fat diet or normal diet. Each µCT scan-based mFE model included either element-specific elastic orthotropic properties calculated from collagen orientation and vTMD (collagen-density model) by experimentally validated formulation, or usual element-specific elastic isotropic material properties dependent on vTMD-only (density-only model). We found that collagen orientation, assessed by circularly polarized light and confocal microscopies, and vTMD, differed among groups and that microindentation results strongly correlate with elastic modulus of collagen-density models (r(2)=0.85, p=10(-5)). Collagen-density models yielded (1) larger strains, and therefore lower strength, in simulations of 3-point bending and physiological loading; and (2) higher correlation between mFE-predicted strength and 3-point bending experimental strength, than density-only models. This novel method supports ongoing translational research to achieve the as yet elusive goal of accurate bone strength prediction.

Understanding the skeletal effects of resistance exercise involves delineating the spatially heterogeneous response of bone to load distributions from different muscle contractions. Bone mineral density (BMD) analyses may obscure these patterns by averaging data from tissues with variable mechanoresponse. To assess the proximal femoral response to resistance exercise, we acquired pretraining and posttraining quantitative computed tomography (QCT) images in 22 subjects (25-55 years, 9 males, 13 females) performing two resistance exercises for 16 weeks. One group (SQDL, n = 7) performed 4 sets each of squats and deadlifts, a second group (ABADD, n = 8) performed 4 sets each of standing hip abductions and adductions, and a third group (COMBO, n = 7) performed two sets each of squat/deadlift and abduction/adduction exercise. Subjects exercised three times weekly, and the load was adjusted each session to maximum effort. We used voxel-based morphometry (VBM) to visualize BMD distributions. Hip strength computations used finite element modeling (FEM) with stance and fall loading conditions. We used QCT analysis for cortical and trabecular BMD, and cortical tissue volume. For muscle size and density, we analyzed the cross-sectional area (CSA) and mean Hounsfield unit (HU) in the hip extensor, flexor, abductor, and adductor muscle groups. Whereas SQDL increased vertebral BMD, femoral neck cortical BMD and volume, and stance hip strength, ABADD increased trochanteric cortical volume. The COMBO group showed no changes in any parameter. VBM showed different effects of ABADD and SQDL exercise, with the former causing focal changes of trochanteric cortical bone, and the latter showing diffuse changes in the femoral neck and head. ABADD exercise increased adductor CSA and HU, whereas SQDL exercise increased the hip extensor CSA and HU. In conclusion, we observed different proximal femoral bone and muscle tissue responses to SQDL and ABADD exercise. This study supports VBM and volumetric QCT (vQCT) to quantify the spatially heterogeneous effects of types of muscle contractions on bone.

Fractures of the proximal femur are the most devastating outcome of osteoporosis. It is generally understood that age-related changes in hip structure confer increased risk, but there have been few explicit comparisons of such changes in healthy subjects to those with hip fracture. In this study, we used quantitative computed tomography and tensor-based morphometry (TBM) to identify three-dimensional internal structural patterns of the proximal femur associated with age and with incident hip fracture. A population-based cohort of 349 women representing a broad age range (21-97years) was included in this study, along with a cohort of 222 older women (mean age 79±7years) with (n=74) and without (n=148) incident hip fracture. Images were spatially normalized to a standardized space, and age- and fracture-specific morphometric features were identified based on statistical maps of shape features described as local changes of bone volume. Morphometric features were visualized as maps of local contractions and expansions, and significance was displayed as Student's t-test statistical maps. Significant age-related changes included local expansions of regions low in volumetric bone mineral density (vBMD) and local contractions of regions high in vBMD. Some significant fracture-related features resembled an accentuated aging process, including local expansion of the superior aspect of the trabecular bone compartment in the femoral neck, with contraction of the adjoining cortical bone. However, other features were observed only in the comparison of hip fracture subjects with age-matched controls including focal contractions of the cortical bone at the superior aspect of the femoral neck, the lateral cortical bone just inferior to the greater trochanter, and the anterior intertrochanteric region. Results of this study support the idea that the spatial distribution of morphometric features is relevant to age-related changes in bone and independent to fracture risk. In women, the identification by TBM of fracture-specific morphometric alterations of the proximal femur, in conjunction with vBMD and clinical risk factors, may improve hip fracture prediction.

Hip fracture risk rises exponentially with age, but there is little knowledge about how fracture-related alterations in hip structure differ from those of aging. We employed computed tomography (CT) imaging to visualize the three-dimensional (3D) spatial distribution of bone mineral density (BMD) in the hip in relation to age and incident hip fracture. We used intersubject image registration to integrate 3D hip CT images into a statistical atlas comprising women aged 21 to 97 years (n = 349) and a group of women with (n = 74) and without (n = 148) incident hip fracture 4 to 7 years after their imaging session. Voxel-based morphometry was used to generate Student's t test statistical maps from the atlas, which indicated regions that were significantly associated with age or with incident hip fracture. Scaling factors derived from intersubject image registration were employed as measures of bone size. BMD comparisons of young, middle-aged, and older American women showed preservation of load-bearing cortical and trabecular structures with aging, whereas extensive bone loss was observed in other trabecular and cortical regions. In contrast, comparisons of older Icelandic fracture women with age-matched controls showed that hip fracture was associated with a global cortical bone deficit, including both the superior cortical margin and the load-bearing inferior cortex. Bone size comparisons showed larger dimensions in older compared to younger American women and in older Icelandic fracture women compared to controls. The results indicate that older Icelandic women who sustain incident hip fracture have a structural phenotype that cannot be described as an accelerated pattern of normal age-related loss. The fracture-related cortical deficit noted in this study may provide a biomarker of increased hip fracture risk that may be translatable to dual-energy X-ray absorptiometry (DXA) and other clinical images.

Purpose: This study aimed to compare peak maximum principal stress in the patellar tendon between persons with and without patellar tendinopathy during a simulated single-leg landing task. A secondary purpose was to determine the biomechanical predictor(s) of peak maximum principal stress in the patellar tendon. Methods: Using finite element (FE) modeling, patellar tendon stress profiles of 28 individuals (14 with patellar tendinopathy and 14 pain-free controls) were created at the time of the peak knee extensor moment during single-leg landing. Input parameters to the FE model included subject-specific knee joint geometry and kinematics, and quadriceps muscle forces. Independent t -tests were used to compare the peak maximum principal stress in the patellar tendon and biomechanical variables used as input variables to the FE model (knee flexion, knee rotation in the frontal and transverse planes and the peak knee extensor moment) between groups. A stepwise regression model was used to determine the biomechanical predictor(s) of peak maximum principal stress in the patellar tendon for both groups combined. Results: Compared with the control group, persons with patellar tendinopathy exhibited greater peak maximum principal stress in the patellar tendon (77.4 ± 25.0 vs 60.6 ± 13.6 MPa, P = 0.04) and greater tibiofemoral joint internal rotation (4.6° ± 4.6° vs 1.1° ± 4.2°, P = 0.04). Transverse plane rotation of the tibiofemoral joint was the best predictor of peak maximum principal stress in the patellar tendon (r = 0.51, P = 0.01). Conclusions: Persons with patellar tendinopathy exhibit greater peak patellar tendon stress compared with pain-free individuals during single-leg landing. The magnitude of peak patellar tendon stress seems to be influenced by the amount of tibiofemoral rotation in the transverse plane. [ABSTRACT FROM AUTHOR]

Hip fractures present a significant healthcare challenge, especially within aging populations, where they are often caused by falls. These fractures lead to substantial morbidity and mortality, emphasizing the need for timely surgical intervention. Despite advancements in medical care, hip fractures impose a significant burden on individuals and healthcare systems. This paper focuses on the prediction of hip fracture risk in older and middle-aged adults, where falls and compromised bone quality are predominant factors. We propose a novel staged model that combines advanced imaging and clinical data to improve predictive performance. By using convolutional neural networks (CNNs) to extract features from hip DXA images, along with clinical variables, shape measurements, and texture features, our method provides a comprehensive framework for assessing fracture risk. The study cohort included 547 patients, with 94 experiencing hip fracture. A staged machine learning-based model was developed using two ensemble models: Ensemble 1 (clinical variables only) and Ensemble 2 (clinical variables and DXA imaging features). This staged approach used uncertainty quantification from Ensemble 1 to decide if DXA features are necessary for further prediction. Ensemble 2 exhibited the highest performance, achieving an Area Under the Curve (AUC) of 0.9541, an accuracy of 0.9195, a sensitivity of 0.8078, and a specificity of 0.9427. The staged model also performed well, with an AUC of 0.8486, an accuracy of 0.8611, a sensitivity of 0.5578, and a specificity of 0.9249, outperforming Ensemble 1, which had an AUC of 0.5549, an accuracy of 0.7239, a sensitivity of 0.1956, and a specificity of 0.8343. Furthermore, the staged model suggested that 54.49% of patients did not require DXA scanning. It effectively balanced accuracy and specificity, offering a robust solution when DXA data acquisition is not always feasible. Statistical tests confirmed significant differences between the models, highlighting the advantages of the advanced modeling strategies. Our staged approach offers a cost-effective holistic view of patients' health. It could identify individuals at risk with a high accuracy but reduce the unnecessary DXA scanning. Our approach has great promise to guide interventions to prevent hip fractures with reduced cost and radiation., Competing Interests: Conflict of Interest Disclosure Statement All authors declare that there are no conflicts of interest.

Aims: Current machine learning-based (ML) models usually attempt to utilize all available patient data to predict patient outcomes while ignoring the associated cost and time for data acquisition. The purpose of this study is to create a multi-stage machine learning model to predict cardiac resynchronization therapy (CRT) response for heart failure (HF) patients. This model exploits uncertainty quantification to recommend additional collection of single-photon emission computed tomography myocardial perfusion imaging (SPECT MPI) variables if baseline clinical variables and features from electrocardiogram (ECG) are not sufficient., Methods: 218 patients who underwent rest-gated SPECT MPI were enrolled in this study. CRT response was defined as an increase in left ventricular ejection fraction (LVEF) > 5% at a 6±1 month follow-up. A multi-stage ML model was created by combining two ensemble models: Ensemble 1 was trained with clinical variables and ECG; Ensemble 2 included Ensemble 1 plus SPECT MPI features. Uncertainty quantification from Ensemble 1 allowed for multi-stage decision-making to determine if the acquisition of SPECT data for a patient is necessary. The performance of the multi-stage model was compared with that of Ensemble models 1 and 2., Results: The response rate for CRT was 55.5% (n = 121) with overall male gender 61.0% (n = 133), an average age of 62.0±11.8, and LVEF of 27.7±11.0. The multi-stage model performed similarly to Ensemble 2 (which utilized the additional SPECT data) with AUC of 0.75 vs. 0.77, accuracy of 0.71 vs. 0.69, sensitivity of 0.70 vs. 0.72, and specificity 0.72 vs. 0.65, respectively. However, the multi-stage model only required SPECT MPI data for 52.7% of the patients across all folds., Conclusions: By using rule-based logic stemming from uncertainty quantification, the multi-stage model was able to reduce the need for additional SPECT MPI data acquisition without sacrificing performance., Competing Interests: Conflict of Interest Disclosure Statement All authors declare that there are no conflicts of interest.