PT  - JOURNAL ARTICLE
AU  - Ferguson, Spencer
AU  - Tille, Patricia M.
TI  - Public Datasets: A Foundation to Artificial Intelligence in Health Care
AID  - 10.29074/ascls.2024003254
DP  - 2024 Dec 05
TA  - American Society for Clinical Laboratory Science
4099  - http://hwmaint.clsjournal.ascls.org/content/early/2024/11/26/ascls.2024003254.short
4100  - http://hwmaint.clsjournal.ascls.org/content/early/2024/11/26/ascls.2024003254.full
AB  - The use of artificial intelligence (AI) in health care is predicated on its safety and efficacy. AI is a technical field of study and is fast evolving. It will affect everyone, so it is important that stakeholders, especially providers and legislators, understand the mechanisms of how AI works so they can make competent decisions to ensure patient safety. There are examples of successful AI systems in health care, but widespread application and adoption suffer due to several issues regarding the type of training data used. All AI systems must be trained using data, and the quality and quantity of this data are at the foundation of their success. Open data addresses issues of validation, reproducibility, and bias within AI systems. Initiatives from private and government agencies, including funding and legislation, support the propagation of open data for research use. The sharing of curated data and trained AI models will exponentially increase AI development in health care. Despite hurdles, open data is the key to implementing safe, reproducible AI models in health care.