Deep Learning Detects Osteoporosis From Medical Imaging
Researchers from Guangzhou University of Traditional Chinese Medicine conducted a systematic review and meta-analysis, searching PubMed, Embase, Cochrane, and Web of Science through May 16, 2024. They included 60 studies totaling 66,195 participants and found high pooled sensitivity and specificity for x-ray and CT–based deep learning models (x-ray appendicular SEN 0.97 SPC 0.90; CT lumbar SEN 0.91 SPC 0.92), but identified few MRI studies and limited external validation.
Key Points
- 1Reports pooled high sensitivity and specificity for x-ray and CT-based deep learning models
- 2Identifies only three MRI studies and pervasive lack of adequate external validation across studies
- 3Recommends external validation and broader datasets to build generalizable AI diagnostic tools for osteoporosis
Scoring Rationale
Comprehensive synthesis and robust pooled metrics, offset by limited MRI evidence and inadequate external validation across studies.
Sources
Public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems
