From X-Rays to Early Intervention: Cost-Effectiveness of Deep Learning-Powered Opportunistic Osteoporosis Screening in an Asian country

Principal Investigator: Mickael Hiligsmann

Presenter: Manju Chandran

Sponsor of the abstract (ASBMR member): Manju Chandran

Abstract Review Categories: Osteoporosis – Assessment

Key words: osteoporosis, screening, artificial intelligence

A silent epidemic with devastating consequences, osteoporosis remains underassessed globally with more than half of at-risk women never formally evaluated or diagnosed. In Asia, this diagnostic gap is exacerbated by limited dual-energy X-ray absorptiometry (DXA) access and fragmented care. While Singapore enjoys broad access to DXA, real-world screening remains suboptimal, and osteoporosis is often missed during routine care. Chest radiographs (CXRs) are among the most frequently performed imaging tests performed world-wide representing a massive untapped method for opportunistic screening. We evaluated the cost-effectiveness of a novel, AI-powered deep learning (DL) solution (PROS® CXR: OSTEO) applied to CXRs for opportunistic osteoporosis screening in Singapore. Employing a validated microsimulation-based Markov model, we compared two strategies in women aged ≥50: (1) DL-enabled opportunistic screening followed by confirmatory DXA and treatment, versus (2) no screening or treatment. The model incorporated real-world data on fracture incidence, healthcare costs, DXA referral rates, DL model performance, treatment uptake and adherence.

At a base-case osteoporosis prevalence of 17%, the DL strategy produced significant clinical gains with 17 fractures prevented, 7 life-years gained, and 16 quality-adjusted life years (QALYs) added per 10,000 women at an incremental cost-effectiveness ratio (ICER) of SGD$ 46,412 (~USD 36,232.91) per QALY gained. This is well below Singapore’s willingness-to-pay threshold of SGD$ 85,000 (~USD 66,357.78). Across a prevalence range (9.3%-37%), the ICER remained favorable (SGD$71,162–SGD$24,749 per QALY). The approach remained  cost-effective if the per-patient AI screening cost was under SGD$ 93 (~USD 72.60).

This is the first cost-effectiveness analysis of AI-enabled chest X-ray–based osteoporosis screening in Asia, demonstrating that this scalable strategy is clinically meaningful and economically viable even in a high-resource setting like Singapore. By embedding fracture risk assessment into existing and easily available imaging modalities, healthcare systems can expand detection without requiring additional imaging infrastructure.

In an era of aging populations and increasing fracture burden, our findings support AI-powered opportunistic screening as a transformative, cost-effective strategy to close the osteoporosis care gap.

Funding Acknowledgement: Promedius

Conflict of Interest:  MC has received lecture fees and travel grants from Amgen and Promedius, JYR has received consulting fees and research grant from Promedius.  MH has received research grants and lecture fees (paid to institution) from Radius Health, Angelini Pharma and IBSA and grant advisory and consultant fees (Paid to Institution) from  Pfizer and Grünenthal.