Artificial Intelligence-Enabled Electrocardiography Identifies Osteoporosis and has Prognostic Value.

Background

Osteoporosis, a common disease leading to weakened bones and increased fracture risk, often goes undiagnosed until a fracture occurs. Dual-energy X-ray absorptiometry (DXA) is the current gold standard for bone mineral density (BMD) measurement, but it has limitations. Recent studies reported Artificial Intelligence (AI)-enabled Electrocardiography (ECG) for disease screening. We hypothesized that AI ECG could serve as a screening tool for osteoporosis.

Objective

This study aimed to develop a deep learning model (DLM) to identify osteoporosis using EKG features and to assess its performance and clinical implications.

Methods

We conducted a retrospective study involving 25,401 patients who underwent 44,732 EKGs with DXA-measured BMD at two hospitals. The area under the receiver operating characteristic curve (AUC) was used for evaluation. Additionally, our DLM was tested for predicting mortality using Kaplan-Meier survival analysis and the Cox proportional hazards model.

Results

The DLM achieved an AUC of 0.741 in internal validation and 0.868 in external validation for detecting osteoporosis. Furthermore, the negative predictive value for osteoporosis was 93.7% in the internal set and 85.8% in the external set. The DLM-detected osteoporosis group exhibited a higher risk of all-cause mortality with a hazard ratio (HR) of 2.06 (95% CI: 1.23–3.45) in the internal validation set, and similar results were observed in the external validation set (HR: 1.87, 95% CI: 1.21–2.89).

Conclusion

Our DLM, utilizing EKG for osteoporosis identification, demonstrated impressive results. It has the potential to serve as a cost-effective and practical screening tool for early osteoporosis detection, with significant prognostic implications.