Skip to content

AI analysis of X-rays predicts response to osteoporosis treatment

Development of a Deep Learning-Based Model for Evaluating and Predicting Osteoporosis Treatment Outcomes Using Plain Radiographs

Status
Active, not recruiting
Phases
Unknown
Study type
Observational
Source
CRIS
Registry ID
KCT0011129
Enrollment
3000
Registered
2025-11-11
Start date
2025-11-24
Completion date
Unknown
Last updated
2025-11-17

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

None listed

Interventions

None listed

Sponsors

Korea University Ansan Hospital
Lead Sponsor

Eligibility

Sex/Gender
All

Inclusion criteria

Inclusion criteria: #. Retrospective Cohort Inclusion Criteria - Adults who underwent bone mineral density (BMD) testing at our institution during the past ~20 years with longitudinal DXA follow-up at intervals =1 year. - Lumbar spine radiographs (AP and lateral) obtained within ±3 months of each DXA time point. #. Prospective Cohort Inclusion Criteria - Patients visiting our institution who meet any of the following: - Women =65 years or men =70 years with osteoporosis or osteopenia meeting FRAX treatment thresholds, confirmed as requiring treatment; - Patients younger than the above age thresholds may be included if investigator assessment and testing confirm a need for treatment. - Patients already on osteoporosis medication may be included if continued therapy for =6 months is planned and the study team deems inclusion aligned with study objectives. - Able to complete =6 months of follow-up from study start. - DXA, blood tests, and lumbar spine radiographs can be obtained both at baseline and at the 6-month follow-up. - The 6-month follow-up window is anchored to the treatment initiation date. - Ability to collect baseline clinical information (sex, age, height, weight, etc.) at enrollment.

Exclusion criteria

Exclusion criteria: #. Retrospective Cohort Exclusion Criteria - Imaging data not available (missing or not archived). - Imaging available but not suitable for analysis, including: - Poor image quality. - History of lumbar spinal fusion/fixation or implants (e.g., instrumentation, cement augmentation) that may confound interpretation. - Pathologic fracture, congenital or severe spinal deformity, or infection that precludes analysis. #. Prospective Cohort Exclusion Criteria - Failure to complete the required assessments at baseline and at the 6-month follow-up. - Note: Follow-up assessments are acceptable if performed from Day 151 (5 months + 1 day) to Day 180 (6 months) after the first study test. - Imaging obtained but unsuitable for analysis, including: - Poor image quality. - History of lumbar spinal fusion/fixation or implants (e.g., instrumentation, cement augmentation) that may confound interpretation. - Pathologic fracture, congenital or severe spinal deformity, or infection that precludes analysis.

Design outcomes

Primary

MeasureTime frame
Percent change in bone turnover marker (CTX, P1NP) at 6 months from baseline

Secondary

MeasureTime frame
Correlation/regression between imaging-based indices and ?CTX/?P1NP;Responder rate (e.g., =25% decrease in CTX or =20% change in P1NP) with AUC/accuracy/sensitivity/specificity

Countries

Korea, Republic of

Contacts

Public ContactHyung Jun Park

Korea University Ansan Hospital

ppakyung@hotmail.com@hotmail.com+82-31-412-5049

Outcome results

None listed

Source: CRIS (via WHO ICTRP) · Data processed: Feb 4, 2026