The Covid-19 pandemic has shed a harsh and sometimes unsustainable light on fragile and painful areas of our society, from emergency services to the shelters for refugees, homeless people or battered women. In the same time, AI has proven to be a formidable tool in the field of governance of health issues... Provided, warns Jay Shaw, the Director of Research on Ethics of AI in Health at the Integrated Centre of Bioethics from the University of Toronto1, to think according to an ethic of the design.
At that time, the research platform on the ethics of AI in health recurrently explored 4 main themes. First, the public confidence, not only in the technology, say AI, but also in the institutions that govern it: do we trust in the government to regulate AI? Second, the future of care. It is the conversation about the future of work applied to health care. For example, how does AI and other disruptive technologies promise to change what is currently at the heart of the practice of health care providers, such as compassionate and centred care on people? Third, equity and bias. This is a growing field of research which has recently addressed the issues of equity, accountability and transparency of AI. We are less interested in the technical aspects than in the adoption of an broader perspective on these issues. What are the inputs that feed the AI? And what are the consequences in the world of such an system? And what constitutes fairness? The last theme is the ethical governance. This multidimensional aspect covers many topics such as access to data sets, but also questions more difficult the role of private actors, especially large companies technology companies that are increasingly present in the health.
An example of a project for which I have submitted a grant application just before the pandemic is the study of values embedded in design of RNs for health care. The idea is that in different types organisations, there will be different values embedded in the choices of design and its implementation. We want to study this hypothesis in three different contexts, all of which design an RN for the care of health: a large technological organisation, a start-up and finally an health care organisation (a hospital). In each context, we want to explore the following questions and compare their different answers: What values are implemented in the design? And then in real-life implementations? This area of research is called design ethics, and the research is called the methodology that we would use is called multi-site ethnography which
is about going into different environments to talk to people, both by getting involved with them and observing what they do, and then at the same time to analyse the qualitative data we have collected.
The pandemic has put an end to all non-essential research projects, by substantially all of our AI research has been discontinued. Firstly, we could no longer conduct interviews or field observations, and secondly, with certain specific projects, we could no longer interrupt the normal provision of health care.
What we are seeing now is that Covid-19 is changing each and every part of the health system: emergency services, violence domestic servants, refugee shelters, ... It has inevitably raised questions of interest to all health researchers. Thus, the call for proposals for research that grants funding receives a lot of support more demands than before the pandemic. Pragmatically, we need to look for other ways to finance our project. What I have also been thinking about recently is that normally the technology development is a long-term process, but the development of a pandemic has shortened all deadlines and requires immediate action.
Many things are urgent now, and we are not sure how much time we have left how this will affect processes in the long term.
We are currently working on 2 projects.
The first deals with the issue of health equity in healthcare particularly raised in the context of the pandemic and virtual increasing use of telemedicine. How could we
improving inclusion in virtual health care? It is clear that the whole the world does not have equal access to a telephone call or to an Internet platform of online video conferencing, not to mention the skills needed to log in. Our literature review identified three different levels for e-inclusion: access to technology itself, the ability to
technology, then the privilege of using the Internet in the most efficient way significant possible. We also found that a more simple reinforces equity. If our application for funding is approved, we would like to study cases of use in four provinces and five U.S. states, to discover strategies for that actually contribute to promoting inclusion in care virtual.
The second project focuses on the persistence of trust in the institutions responsible for using health data to inform the public about the policies. Indeed, while we bring together large sets of data to feed AI in unprecedented quantities, how to feed it in a way that has never been done before can we ensure that trust in the institutions is maintained responsible for the governance of health data? In Ontario, the of data dedicated to mitigating the threat of pandemic by AI, and other data analysis (e.g. to model propagation), a was the first to collect so much sensitive data. How ensure that these initiatives maintain public confidence in institutions and do not meet with strong resistance from the public? We can start with a few good summaries that describe the current use of AI during the pandemic (prediction of the deterioration once the patient is admitted to the intensive care units, support for the diagnosis of Covid-19 based on lung X-rays, for admission to hospital) and also analyse the data sets used to feed AI. It is a way to Understand concretely how to use AI technologies with integrity and thus deserve the public's trust.
Interview by Lauriane Gorce
.png)
.png)
