Reporting, Assessing Risk of Bias in Artificial Intelligence Models for Healthcare Research
Updated: Jul 18, 2021
Gary Collins et al. write in BMJ Open (2021) concerning guidelines and reporting protocols addressing the risk of bias in AI/ML, utilized in healthcare research, particularly in diagnostic and prognostic applications.
The article entitled, "Protocol for development of a reporting guideline (TRIPOD-AI) and risk of bias tool (PROBAST-AI) for diagnostic and prognostic prediction model studies based on artificial intelligence," (Collins, et al., 2021) provides exhaustive background information concerning AI/ML, biases in training data, and future directions for evaluating, reporting, and assessing the risk of bias.
According to the authors, "Both TRIPOD-AI and PROBAST-AI will be developed following published guidance from the EQUATOR Network" (ibid.). Moreover, the authors envision a coordinated effort which will result in a formalized study protocol, a systemic review protocol, and explanatory papers (ibid.).
James Burroughs, CEO of North Carolina Health Equity Project, Inc., said in his weekly call with stakeholders, "North Carolina Health Equity Project, Inc. is keenly aware of biases in AI/ML, promulgated by human interests, and incomplete, or even blatantly biased training data which further amplify bias. We are eager to see further recommendations and guidelines from collaborators on TRIPOD-AI and PROBAST-AI, particularly as we innovate for all people, improving health outcomes and saving lives." Readers can learn more about the commitment of North Carolina Health Equity Project, Inc. to safe and ethical innovation at Our Commitment to Safe and Ethical Innovation in Public Health R&D (nchealtheq.org).
Read the full article at Protocol for development of a reporting guideline (TRIPOD-AI) and risk of bias tool (PROBAST-AI) for diagnostic and prognostic prediction model studies based on artificial intelligence | BMJ Open.
Part of health disparities news and research reporting by North Carolina Health Equity Project, Inc. Image Credit: Lagos Techie on Unsplash. The views expressed by the authors of any cited publication(s) or external link(s) do not necessarily reflect the views of North Carolina Health Equity Project, Inc., nor are any affiliations or endorsements implied.
The Mission and Vision of North Carolina Health Equity Project, Inc.
North Carolina Health Equity Project, Inc., NCHEALTHEQ™, develops adaptive, intelligent, next-generation technologies to improve health outcomes for all people, especially those in marginalized communities. We leverage the power of cross-disciplinary innovation and disruptive technologies including AI/ML, “accelerating research and saving lives.” Learn more at https://www.nchealtheq.org/.
Corporate and Governmental Affairs
North Carolina Health Equity Project, Inc.