A review of deep learning in orthopaedics
Recent developments in artificial intelligence have drawn considerable interest in its medical applications. Dr Laith Alzubaidi led an extensive 40-page review covering the applications, challenges, trustworthiness and fusion of artificial intelligence in orthopaedics and in medicine more broadly. It was published in Artificial Intelligence in Medicine on 25 July 2024.
Why trustworthiness is the hard part
In orthopaedics the obvious applications of deep learning are in imaging: detecting fractures, classifying patterns of wear, measuring angles and positions that are currently measured by hand. The accuracy of such systems on test data is, however, the easier problem. The harder questions are the ones the review addresses under trustworthiness: whether a model's output can be explained, how it behaves on data unlike anything it was trained on, and who is accountable when it is wrong. These are the questions that determine whether such tools can responsibly be used in clinical decisions.
It is a valuable resource for anyone interested in the application of artificial intelligence in orthopaedics.
About the lead author
Dr Laith Alzubaidi is a postdoctoral researcher at Queensland University of Technology. He holds an MSc in computer science from the University of Missouri (2016), has been awarded two scholarships, and has published more than 40 refereed papers, including more than 25 joint publications.
Read it
The review is available on ScienceDirect.
Related information
See also our post on a deep learning framework for detecting abnormalities in shoulder X-ray images and on the future of shoulder range of motion assessment.