Ehealth

'When digital health meets digital capitalism, how many common goods are at stake?' by Tamar Sharon in (2018) Big Data and Society 1–12 comments

In recent years, all major consumer technology corporations have moved into the domain of health research. This ‘Googlization of health research’ (‘GHR’) begs the question of how the common good will be served in this research. As critical data scholars contend, such phenomena must be situated within the political economy of digital capitalism in order to foreground the question of public interest and the common good. Here, trends like GHR are framed within a double, incommensurable logic, where private gain and economic value are pitted against public good and societal value. While helpful for highlighting the exploitative potential of digital capitalism, this framing is limiting, insofar as it acknowledges only one conception of the common good. This article uses the analytical framework of modes of justification developed by Boltanksi and Thevenot to identify a plurality of orders of worth and conceptualizations of the common good at work in GHR. Not just the ‘civic’ (doing good for society) and ‘market’ (enhancing wealth creation) orders, but also an ‘industrial’ (increasing efficiency), a ‘project’ (innovation and experimentation), and what I call a ‘vitalist’ (proliferating life) order. Using promotional material of GHR initiatives and preliminary interviews with participants in GHR projects, I ask what moral orientations guide different actors in GHR. Engaging seriously with these different conceptions of the common good is paramount. First, in order to critically evaluate them and explicate what is at stake in the move towards GHR, and ultimately, in order to develop viable governance solutions that ensure strong ‘civic’ components.

Sharon argues

In the last few years, every major consumer technology corporation, from Google to Apple, to Facebook, Amazon, Microsoft and IBM, has moved decisively into the health and biomedical sector. These are companies that, for the most part, have had little interest in health in the past, but that by virtue of their data expertise and the large amounts of data they already have access to, are becoming important facilitators, if not initiators, of data-driven health research and healthcare. 

This ‘Googlization of health research’ (GHR), as I have called this process elsewhere (Sharon, 2016), promises to advance health research by providing the technological means for collecting, managing and analysing the vast and heterogeneous types of data required for data-intensive personalized and precision medicine. Apple’s ResearchKit software, for example, which turns the iPhone into a platform for conducting medical studies, allows researchers to access diverse types of data (sleeping patterns, food consumption, gait), to recruit larger numbers of participants than average in clinical trials, and to monitor participants in real time (Savage, 2015). Similarly, the new analytics techniques and data repositories offered by consumer technology companies seek to overcome limitations in traditional medical analytics methods and infrastructure. DeepMind, for example, Google’s London-based artificial intelligence offshoot, is applying deep learning for the prediction of cardiovascular risk, eye disease, breast cancer and patient outcomes, in collaboration with several hospitals (Poplin et al., 2018; Ram, 2018). Verily, Alphabet’s life science branch, is developing new tools to capture and organize unstructured health data, for example in its ‘Project Baseline’ in partnership with Stanford and Duke University. The study will collect and analyse a wide range of genetic, clinical and lifestyle data on 10,000 healthy volunteers, with the aim of comprehensively ‘mapping human health’ (Verily, 2018). Google, Microsoft, Amazon and IBM have also begun packaging their clouds as centralized genomic databases where researchers can store and run queries on genomic data. 

Many of these techniques still have not delivered on their promises, all the while introducing a host of new challenges and limitations, such as new selection and other types of biases (Agniel et al., 2018; Hemkens et al., 2016; Jardine et al., 2015). Yet their potential, if not over-hyped, remains promising (Fogel et al., 2018), and places these corporations in a privileged position in the move towards personalized medicine and Big Data analytics – and broader healthcare vistas. Indeed, most recently a number of these companies have begun moving into the domains of electronic health record management, employee healthcare and health insurance (Farr, 2017; Farr, 2018; Wingfield et al., 2018). 

Beyond these promises, GHR also raises a number of challenges and risks. First amongst these are concerns of privacy and informed consent. GHR is an instance of data-intensive research characterized by the use of large digital datasets and Big Data analytics, where traditional mechanisms put in place to protect research participants are increasingly under strain. These issues may be exacerbated in situations where consumer technology companies, whose data-sharing practices often are not subject to the same privacy-protecting regulations and codes of conduct as those of medical researchers, are involved (Zang et al., 2015). The potential for ‘context transgressions’ (Nissenbaum, 2010), whereby data may flow between medical, social and commercial contexts governed by different privacy norms, is greater here. Furthermore, broader questions about the value of personal health data and publicly generated datasets, and what market advantage is conferred to commercial entities who can access them and develop treatments and services based on this access, will emerge. In other words, in GHR initiatives, concerns that are common in the practices of digital capitalism are imported into the health realm (Sharon, 2016). 

A recent controversy surrounding a data sharing partnership between Google DeepMind and the NHS illustrates how some of these issues are already playing out. Announced in 2016, the collaboration between DeepMind and the Royal Free London, a NHS Foundation Trust, granted DeepMind access to identifiable information on 1.6 million of its patients in order to develop an app to help medical professionals identify patients at risk of acute kidney injury (AKI). The terms of this agreement have been analysed in depth by Powles and Hodson (2017, 2018), who argue that it lacked transparency and suffered from an inadequate legal and ethical basis. Indeed, following an investigation, the Information Commissioner’s Office (ICO, 2017) ruled that this transfer of data and its use for testing the app breached data protection law. Namely, patients were not at all aware that their data was being used. Under UK common law, patient data can be used without consent if it is for the treatment of the patient, a principle known as ‘direct care’, which the Trust invoked in its defence. But as critics argue, insofar as only a small minority of the patients whose data was transferred to DeepMind had ever been tested or treated for AKI, appealing to direct care could not justify the breadth of the data transfer. 

Of course, GHR collaborations taking place in different jurisdictions will be provided with different opportunities and face different legal challenges. And despite the global profile of the corporations in question, national and regional guidelines for the management of AI and Big Data in health will impact what GHR collaborations can and cannot do. But the DeepMind case also raises questions beyond data protection, privacy and informed consent, which have to do with the newfound role that tech corporations will play in health research and healthcare, and new power asymmetries between corporations, public health institutions and citizens that may ensue. For example, will these corporations become the gatekeepers of valuable health datasets? What new biases may be introduced into research using technologies, such as iPhones, that only certain socio-economic segments of the population use? What role will these companies, already dominant in other important domains of our lives, begin to play in setting healthcare agendas? These are questions that concern collective and societal benefit – broadly  speaking, the common good. They point to the need to situate the analysis of GHR in the wider context of the political economy of data sharing and use, and they foreground a number of concerns that move beyond (just) privacy and informed consent, including social justice, accountability, democratic control and the public interest. 

These values are the focus of the growing body of literature in critical data studies that draws on a political economy critique to address the development of new power asymmetries and discriminations emerging in Big Data infrastructures (Taylor, 2017; van Dijk, 2014; Zuboff, 2015). In this context, new Big Data divides can be expected based on access to and ownership of data, technological infrastructures and technical expertise, with important repercussions for who shapes the future of (health) research (boyd and Crawford, 2012). However, by focusing on the new power asymmetries emerging between data subjects and corporations, critical data studies tend to frame data sharing in terms of two incommensurable logics: public benefit and private, corporate gain. In this article, I argue that this dichotomy is limiting, insofar as it only allows for one vision of the common good, while a plurality of conceptualizations of the common good are at work in GHR. In the following, I use the interpretive framework of economies of worth developed by the sociologists Luc Boltanski and Laurent Thevenot (2006 [1991]) to identify a number of moral repertoires that each draw upon different conceptualizations of the common good and that are mobilized by actors in GHR-type initiatives. Doing so depicts a much richer ethical terrain of GHR than is accounted for in most critical analyses of digital capitalism. 

This is valuable for several reasons. First, it is paramount that the moral orientations of actors in GHR be taken seriously, insofar as they influence and guide decision-making processes that are currently taking place. Here I draw on the constructivist tradition that views the discourses, repertoires and logics that convey moral orientations as performative; as contributing to the enactment of technological futures (Foucault, 1965; Latour and Woolgar, 1979). Critical research on GHR must engage with these competing moral orientations and conceptualizations of the common good. Second, this type of mapping is a necessary first step towards critically evaluating different moral repertoires, insofar as it contributes to rendering explicit the trade-offs that will be involved in the enactment of different repertoires. In the current situation, where no comprehensive ethical and policy guidance for GHR exists, this is required if we are to have serious public deliberation about what is at stake in the move towards GHR. Finally, while Boltanski and The´venot’s framework was developed as a descriptive project, I argue that it can be used to help develop normative guidelines for governance of GHR-type projects, and that this should be further developed into a research programme. Here, solutions can be thought of as combinations of repertoires, where different repertoires can check and balance each other. Such solutions will have a good chance of adoption insofar as they will appeal to a wide range of actors. Further, if what Boltanski and The´venot call the ‘civic’ order of worth embodies the most publicly legitimate conception of the common good, we can design solutions that ensure the presence of strong civic components. For this, however, the civic repertoire must be ‘updated’, so to speak: it must first engage seriously with competing conceptions of the common good that are mobilized in the empirical reality of GHR. The article thus seeks to map and analyse the different orders of worth invoked by actors involved in GHR as a first step towards this endeavour.