The increasing implementation of and reliance on machine-learning (ML) algorithms to perform tasks, deliver services and make decisions in health and healthcare have made the need for fairness in ML, and more specifically in healthcare ML algorithms (HMLA), a very important and urgent task. However, while the debate on fairness in the ethics of artificial intelligence (AI) and in HMLA has grown significantly over the last decade, the very concept of fairness as an ethical value has not yet been sufficiently (...) explored. Our paper aims to fill this gap and address the AI ethics principle of fairness from a conceptual standpoint, drawing insights from accounts of fairness elaborated in moral philosophy and using them to conceptualise fairness as an ethical value and to redefine fairness in HMLA accordingly. To achieve our goal, following a first section aimed at clarifying the background, methodology and structure of the paper, in the second section, we provide an overview of the discussion of the AI ethics principle of fairness in HMLA and show that the concept of fairness underlying this debate is framed in purely distributive terms and overlaps with non-discrimination, which is defined in turn as the absence of biases. After showing that this framing is inadequate, in the third section, we pursue an ethical inquiry into the concept of fairness and argue that fairness ought to be conceived of as an ethical value. Following a clarification of the relationship between fairness and non-discrimination, we show that the two do not overlap and that fairness requires much more than just non-discrimination. Moreover, we highlight that fairness not only has a distributive but also a socio-relational dimension. Finally, we pinpoint the constitutive components of fairness. In doing so, we base our arguments on a renewed reflection on the concept of respect, which goes beyond the idea of equal respect to include respect for individual persons. In the fourth section, we analyse the implications of our conceptual redefinition of fairness as an ethical value in the discussion of fairness in HMLA. Here, we claim that fairness requires more than non-discrimination and the absence of biases as well as more than just distribution; it needs to ensure that HMLA respects persons both as persons and as particular individuals. Finally, in the fifth section, we sketch some broader implications and show how our inquiry can contribute to making HMLA and, more generally, AI promote the social good and a fairer society. (shrink)

Biomedical research using data from participants’ mobile devices borrows heavily from the ethos of the “citizen science” movement, by delegating data collection and transmission to its volunteer subjects. This engagement gives volunteers the opportunity to feel like partners in the research and retain a reassuring sense of control over their participation. These virtues, in turn, give both grass-roots citizen science initiatives and institutionally sponsored mHealth studies appealing features to flag in recruiting participants from the public. But while grass-roots citizen science (...) projects are often community-based, mHealth research ultimately depends on the individuals who own and use mobile devices. This inflects the ethos of mHealth research towards a celebration of individual autonomy and empowerment, at the expense of its implications for the communities or groups to which its individual participants belong. But the prospects of group harms — and benefits — from mHealth research are as vivid as they are in other forms of data-intensive “precision health” research, and will be important to consider in the design of any studies using this approach. (shrink)

Surveillance is essential for communicable disease prevention and control. Traditional notification of demographic and clinical information, about individuals with selected infectious diseases, allows appropriate public health action and is protected by public health and privacy legislation, but is slow and insensitive. Big data–based electronic surveillance, by commercial bodies and government agencies, which draws on a plethora of internet- and mobile device–based sources, has been widely accepted, if not universally welcomed. Similar anonymous digital sources also contain syndromic information, which can be (...) analysed, using customised algorithms, to rapidly predict infectious disease outbreaks, but the data are nonspecific and predictions sometimes misleading. However, public health authorities could use these online sources, in combination with de-identified personal health data, to provide more accurate and earlier warning of infectious disease events—including exotic or emerging infections—even before the cause is confirmed, and allow more timely public health intervention. Achieving optimal benefits would require access to selected data from personal electronic health and laboratory records and the potential to re-identify individuals found to be involved in outbreaks, to ensure appropriate care and infection control. Despite existing widespread digital surveillance and major potential community benefits of extending its use to communicable disease control, there is considerable public disquiet about allowing public health authorities access to personal health data. Informed public discussion, greater transparency and an ethical framework will be essential to build public trust in the use of new technology for communicable disease control. (shrink)

Background Outbreaks of infectious disease cause serious and costly health and social problems. Two new technologies – pathogen whole genome sequencing and Big Data analytics – promise to improve our capacity to detect and control outbreaks earlier, saving lives and resources. However, routinely using these technologies to capture more detailed and specific personal information could be perceived as intrusive and a threat to privacy. Method Four community juries were convened in two demographically different Sydney municipalities and two regional cities in (...) New South Wales, Australia to elicit the views of well-informed community members on the acceptability and legitimacy of: making pathogen WGS and linked administrative data available for public health researchusing this information in concert with data linkage and machine learning to enhance communicable disease surveillance systems Fifty participants of diverse backgrounds, mixed genders and ages were recruited by random-digit-dialling and topic-blinded social-media advertising. Each jury was presented with balanced factual evidence supporting different expert perspectives on the potential benefits and costs of technologically enhanced public health research and communicable disease surveillance and given the opportunity to question experts. Results Almost all jurors supported data linkage and WGS on routinely collected patient isolates for the purposes of public health research, provided standard de-identification practices were applied. However, allowing this information to be operationalised as a syndromic surveillance system was highly contentious with three juries voting in favour, and one against by narrow margins. For those in favour, support depended on several conditions related to system oversight and security being met. Those against were concerned about loss of privacy and did not trust Australian governments to run secure and effective systems. Conclusions Participants across all four events strongly supported the introduction of data linkage and pathogenomics to public health research under current research governance structures. Combining pathogen WGS with event-based data surveillance systems, however, is likely to be controversial because of a lack of public trust, even when the potential public health benefits are clear. Any suggestion of private sector involvement or commercialisation of WGS or surveillance data was unanimously rejected. (shrink)