When a person suffers a severe break in their thigh bone, known as a femur fracture, quick and accurate diagnosis is vital for proper care. New data shows that artificial intelligence (AI) is highly skilled at spotting these specific types of fractures in medical images.
Researchers looked at over 95,000 femur images across 37 different studies. They compared how well AI performed against both expert and non-expert human readers. The results showed that AI achieved high sensitivity and specificity, meaning it was very effective at correctly identifying the fractures while also correctly ruling out healthy bones.
While experts are already very accurate, the data suggests that AI can also help non-experts perform better. By using AI as a tool, non-expert readers saw an increase in their ability to spot fractures. This suggests that AI could be a helpful teammate in busy medical settings to ensure no fracture goes unnoticed.
Common questions
How accurate is artificial intelligence at finding broken thigh bones?
Artificial intelligence showed high accuracy in identifying acute traumatic femur fractures. It achieved a sensitivity of 0.94 and a specificity of 0.94. These numbers indicate that the technology is very effective at correctly identifying fractures and correctly identifying healthy bones in medical images.
Can artificial intelligence help non-expert staff identify fractures?
Yes, the data suggests that artificial intelligence has potential for augmenting non-expert humans. When non-expert readers used AI assistance, their area under the receiver operating characteristic curve—a measure of diagnostic accuracy—increased from 0.95 to 0.97.
How does AI compare to human experts in finding fractures?
Both experts and AI showed very high accuracy. Experts had an area under the receiver operating characteristic curve of 0.989 when working alone. AI showed a similar high score of 0.98. When experts used AI, their accuracy score reached 0.99.
