Logo image
Back
The ethical considerations including inclusion and biases, data protection, and proper implementation among AI in radiology and potential implications
Journal article   Open access   Peer reviewed

The ethical considerations including inclusion and biases, data protection, and proper implementation among AI in radiology and potential implications

Clarissa Martin, Kyle DeStefano, Harry Haran, Sydney Zink, Jennifer Dai, Danial Ahmed, Abrahim Razzak, Keldon Lin, Ann Kogler, Joseph Waller, …
Intelligence-based medicine, v 6, 100073
2022
DOI: https://doi.org/10.1016/j.ibmed.2022.100073
url
https://doi.org/10.1016/j.ibmed.2022.100073View
Published, Version of Record (VoR) Open

Abstract

Artificial intelligence Ethics Medical Imaging Radiology
Artificial intelligence (AI) has an enorgmoust potential to improve radiologic diagnostics, efficiency, and consistency, through implementation of evidence-based regulations is crucial to prevent ethical violations pertaining to patient privacy, data management, and diagnostic accuracy. Several key terms were searched on PubMed such as “Artificial Intelligence”, "COVID-19", “Ethics” and “Radiology”. Ultimately, 32 articles that were published from June 2016 - November 2020 were included in our review given their presentation of original information and relevance to clinical AI imaging and ethics. Several themes appeared and allowed for the following recommendations: 1. Protocols explaining how patient data will be utilized should be provided to and implemented by providers. 2. Differences in data regulations among industry partners must be considered for successful implementation of AI in radiology, and protections against entities with a financial interest in patient data must be included. 3. Providers should receive education and training on basic AI algorithm methodology to allow them to explain risks to patients and obtain informed consent. 4. Radiologists and AI programmers should work together to allow for feedback, collaboration, and the development of more clinically impactful and ethical AI applications. We conclude by recommending collaboration between providers and programmers with a focus on existing AI platforms over developing new algorithms, as there is a lack of consensus on the proper applications of and regulations needed for already existing technologies. Future research should examine and aim to addressimplicit biases in AI technologies. •Implementing AI in healthcare must ensure physicians evaluate the clinical utility and accuracy of AI systems in practice.•Care must be taken to not excacerbate socieconomic, racial and gender biases in AI algorithms.•Data regulations must make clear who owns, has access, shares, and monitors the patient data to ensure privacy is protected.

Metrics

1 Record Views

Details

Logo image