AI in Diagnosing and Predicting Vertebral Fractures
Author Information
Author(s): Namireddy Srikar R., Gill Saran S., Peerbhai Amaan, Kamath Abith G., Ramsay Daniele S. C., Ponniah Hariharan Subbiah, Salih Ahmed, Jankovic Dragan, Kalasauskas Darius, Neuhoff Jonathan, Kramer Andreas, Russo Salvatore, Thavarajasingam Santhosh G.
Primary Institution: Imperial College London
Hypothesis
Is AI an effective and accurate tool for predicting and diagnosing vertebral fractures?
Conclusion
AI technologies show considerable promise in improving the diagnosis and prognostication of vertebral fractures with high accuracy.
Supporting Evidence
- AI showed high accuracy in diagnosing vertebral fractures with predictive AUROC of 0.82.
- Osteoporotic vertebral fracture diagnosis AUROC was 0.92.
- Non-pathological vertebral fracture diagnosis AUROC was 0.85.
- Vertebral compression fracture diagnosis AUROC was 0.87.
- Traditional models had the highest median AUROC of 0.90 for fracture prediction.
- Deep learning models excelled in diagnosing all fracture types.
- High heterogeneity (I² > 99%) indicated significant variation in model design and performance.
- Further research is needed to standardize AI models and validate them across diverse datasets.
Takeaway
This study shows that computers can help doctors find broken bones in the spine more accurately and quickly.
Methodology
A systematic review and meta-analysis of studies using AI for vertebral fracture diagnosis or prognosis.
Potential Biases
Potential selection and sampling bias in the studies analyzed.
Limitations
High heterogeneity and risk of bias in included studies.
Participant Demographics
Studies included participants aged 50 and older, with varying sample sizes.
Statistical Information
P-Value
p<0.001
Confidence Interval
[0.78–0.85]
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website