Generative Artificial Intelligence Considerations Beyond Natural Language Processing: Computer Vision Applications in Health Care.
Authors
Affiliations (3)
Affiliations (3)
- Midwest Orthopaedics at Rush, Chicago, Illinois, U.S.A.
- Commons Clinic, Long Beach, California, U.S.A.
- Department of Orthopaedic Surgery, Harvard Medical School, Massachusetts General Hospital, Boston, Massachusetts, U.S.A.
Abstract
Generative artificial intelligence describes methodological processes wherein the output of interest for a given dataset is unlabeled (or unknown) and therefore is an original representation of learned data. Beyond natural language processing, computer vision embodies an important technique leveraging generative artificial intelligence methods and has enabled novel use cases relevant to health care. Namely, the introduction of autoencoders, generalized adversarial networks, and diffusion models have propelled the field of medical image reconstruction. Generative imaging reconstruction has implications including, but not limited to, improvements in image quality, surgical planning, and registry augmentation in a risk-averse manner as it pertains to patient privacy concerns. Overall, generative artificial intelligence in health care can enhance the efficiency and quality of medical services and research; however, responsible application requires governance and regulatory frameworks for responsible and ethical use. LEVEL OF EVIDENCE: Level V, expert opinion.