Artificial/Machine Intelligence (A/MI) is rapidly becoming an important data science-based analytic tool across biomedical discovery, clinical research, medical diagnostics and devices, and precision medicine. Such tools and systems can uncover new possibilities for researchers, physicians, and patients, allowing them to make more informed decisions and achieve better outcomes. When deployed, these systems have the potential to enhance efficiency of the health research and care system.
Machine Intelligence in Healthcare: Perspectives on Trustworthiness, Explainability, Usability and Transparency
In the context of this workshop, MI was defined as the ability of a trained computer system to provide rational, unbiased guidance to humans in such a way that achieves optimal outcomes in a range of environments and circumstances. With such promise for applications of MI tools in the healthcare system, the challenges are: how do we trust that what the computer tells us is correct when we don’t understand how it arrived at the output/answer? How do we ensure that these outputs are safe and beneficial for human health? And, if we change the data or environment, how does this affect the output? These questions are especially relevant to clinical care decision making – are the risks of using such tools understood and how can the technology be deployed for maximal benefit?
This workshop was sponsored by the National Institutes of Health (NIH) National Center for Advancing Translational Sciences (NCATS) and organized jointly with the National Cancer Institute (NCI) and the National Institute of Biomedical Imaging and Bioengineering (NIBIB).
To provide experts and the community the opportunity to share their perspectives on current issues associated with incorporation of MI systems into healthcare settings. Meeting outputs were used to develop a white paper on translating MI for clinical applications and the associated process improvement needed when implementing MI systems in healthcare environments.