This artificial intelligence, ethics and governance roundup has been sourced by AMS Phoenix Fellow Dr. Jay Shaw . Dr. Shaw’s Fellowship is focused on understanding the Implications of Artificial Intelligence and Related Technologies for Healthcare Policy and Practice
Artificial Intelligence, Ethics and Governance Roundup
Artificial intelligence (AI) is a general purpose technology, which means it can be used in a huge variety of ways. At the root of applications of AI are massive volumes of data, which are used to train algorithms to achieve particular kinds of outputs. In health care, this means data about the health of individual people and the health of populations.
One way that we can think of the outputs of AI is in terms of two broad categories of use: “decision support” (where some kind of advanced analytics leads to more accurate predictions that can inform human decisions), and “automation” (where the predictive value of algorithms is directly connected to the actual execution of a task). Based on these use cases, AI offers the potential to transform many of the tasks that make up health care, raising important questions about how those transformations will be governed.
The ethical governance of AI in health care is important for many reasons. When a technology with the potential to lead to such drastic change emerges, our current ways of doing things are thrown into question. How accurate are our diagnostic strategies, and could they be better? How might new technologies interfere with the abilities of health care providers to demonstrate compassion when working directly with patients? How will the health system determine which risks are tolerable and which aren’t when deploying AI technologies?
The broad community of researchers who study AI as an emerging technology in health care are beginning to grapple with these issues. Questions about how to introduce AI into health care, and how to govern the resulting transformation, are rich and diverse. The articles presented here provide a sampling of the issues that arise in response to innovations that relate to the diffusion AI in health care (including the management of big data). In Canada we have a rich ecosystem of people paying attention to these issues too, including the Joint Centre for Bioethics at the University of Toronto. So, happy reading, and stay tuned for more deep thinking about this topic here at home.
From What to How: An Overview of AI Ethics Tools, Methods and Research to Translate Principles into Practices
An international synthesis of guidelines on ethics and Artificial intelligence.
An interesting paper from a technical perspective on how explainability in AI enhances implementability… with a health care focus!
On big data, precision medicine, and private insurance.
Ethics and epistemology of big data research. Includes specific suggestions for Research Ethics and Institutional Review Boards.
Canadian Association of Radiologists White Paper on Ethical and Legal Issues Related to Artificial Intelligence in Radiology
Position statement of the The Canadian Association of Radiologists on ethical and legal issues of AI in radiology. Nice paper outlining advanced thinking on this.
Identifying the link between current state of digital health and future state of health-related AI.
Machine Learning and AI Research for Patient Benefit: 20 Critical Questions on Transparency, Replicability, Ethics and Effectiveness
A step toward “best practices” for AI applications in health.
Comparing policy approaches in the USA, UK and EU.
Analysis of the care.data initiative in the UK, where health data was shared to fuel economic growth.
Discussion of the insidious ways that our data is used to “nudge” our actions: Not directly about health, but clear implications.
International standards for evaluating machine learning. It’s technical, and worth digging into.
On the Royal Free Hospital and Deep Mind Health data sharing “deal”.
Dangers of the Digital Fit: Rethinking Seamlessness and Social Sustainability in Data-intensive Healthcare
Denmark experience: data sharing halted because of “accumulation of purposes”.