MétaCan
Menu
Back to cohort
Record W4408945665 · doi:10.1101/2025.03.25.25324645

Spine age estimation using deep learning in lateral spine radiographs and DXA VFA to predict incident fracture and mortality

2025· preprint· en· W4408945665 on OpenAlexaff
Sang Wouk Cho, Namki Hong, Kyoung Min Kim, Young Han Lee, Chang Oh Kim, Hyeon Chang Kim, Yumie Rhee, Brian H. Chen, William D. Leslie, Steven R. Cummings

Bibliographic record

VenuemedRxiv · 2025
Typepreprint
Languageen
FieldEngineering
TopicMedical Imaging and Analysis
Canadian institutionsUniversity of Manitoba
FundersNational Research Foundation of KoreaMinistry of Science and ICT, South KoreaKorea Health Industry Development InstituteNational Research FoundationKorea Disease Control and Prevention Agency
KeywordsSPINE (molecular biology)RadiographyFracture (geology)EstimationOrthodonticsMedicineGeologyRadiologyBiologyEngineeringPaleontologyBioinformatics

Abstract

fetched live from OpenAlex

Abstract Background Spine age estimated from lateral spine radiographs and DXA vertebral fracture assessments (VFAs) could be associated with fracture and mortality risk. Methods In the VERTE-X cohort (n=10,341, age 40 or older; derivation set) and KURE cohort (n=3,517; age 65 or older; external test set), predicted age difference was defined as estimated spine age minus chronological age. The primary outcome was incident fracture. Secondary outcomes included morphologic vertebral fracture, osteoporosis, and incident mortality. Results Incidence of overall fracture was 20.5/1000 and 21.0/1000 person-years (median follow-up 5.4 and 6.6 years) in VERTE-X and KURE, respectively. Spine age discriminated prevalent vertebral fractures and osteoporosis better than chronological age. Higher predicted age difference (PAD) was associated with greater risk of overall (VERTE-x: adjusted HR [aHR] 1.71; KURE: aHR 1.22 per 1 standard deviation [SD] increment), vertebral (aHR 1.55 and 1.34), and non-vertebral fractures (aHR 1.89 and 1.15, p<0.05 for all), independent of chronological age and prevalent vertebral fracture. FRAX hip fracture probabilities based on spine age improved discrimination for incident hip fracture over chronological age (AUROC 0.83 vs. 0.78, p=0.027). Shorter height, lower femoral neck BMD, diabetes, vertebral fractures, and surgical prosthesis were associated with higher predicted age difference, explaining 40% of variance. In the external test set, higher predicted age difference was associated with greater risk of mortality (aHR 1.31 per 1 SD increment, p=0.001), independent of covariates. Conclusion Spine age estimated from lateral spine radiographs and DXA VFA enhanced fracture risk assessment and mortality prediction in adults. Key Points Spine age estimated from lateral spine radiographs and DXA VFA using deep learning outperformed chronological age in discriminating morphologic vertebral fracture and osteoporosis. Higher predicted age difference (predicted spine age minus chronological age) was associated with greater risk of overall, vertebral, and non-vertebral incident fracture, independent of covariates. Male sex, lower height, lower femoral neck BMD, diabetes mellitus, morphologic vertebral fractures, and surgical prosthesis were correlated with higher predicted age difference, explaining up to 40% variance. Higher predicted age difference was associated with greater risk of mortality, independent of chronological age, sex, prevalent morphologic vertebral, fracture, and clinical biomarkers related to mortality including serum albumin, hemoglobin, and creatinine.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.004
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: Observational
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.012
Threshold uncertainty score0.023

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0010.004
Meta-epidemiology (narrow)0.0010.000
Meta-epidemiology (broad)0.0010.001
Bibliometrics0.0010.000
Science and technology studies0.0000.000
Scholarly communication0.0010.000
Open science0.0010.001
Research integrity0.0010.001
Insufficient payload (model declined to judge)0.0010.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.012
GPT teacher head0.275
Teacher spread0.263 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designObservational
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2025
Admission routes1
Has abstractyes

Explore more

Same venuemedRxivSame topicMedical Imaging and AnalysisFrench-language works237,207