On New Year’s Eve 2020, the Canadian start-up BlueDot detected an unusual pneumonia outbreak in Wuhan thanks to its use of advanced data analytics solutions. Only six days later, the World Health Organisation would issue its first public warning about the outbreak of a novel coronavirus, now known as COVID-19. By exposing vulnerabilities in our healthcare systems and demonstrating how digital technologies can help, the COVID-19 pandemic has provided a powerful impetus to accelerate the digitalisation of healthcare. From improved early warning tools to enhanced diagnosis capacities and personalised treatments, there are few domains in which Artificial Intelligence (AI) and digital technologies have as much potential as in healthcare. Yet, there are also very few sectors that embody so well the significant risks and challenges that we face in harnessing digital innovation. Furthermore, on-the-ground experience during the crisis has also underscored that we still have ground to cover before this potential becomes reality.
In a number of countries, in Asia in particular, digital solutions based on biometrics and geo-localisation data have played a critical role in containing the virus by helping authorities diagnose COVID-19 patients, trace-back their movements over time, and establish virus transmissions chains. In this way, they made clear that it can be possible to “flatten the curve” without resorting to blanket mobility restrictions and costly lockdowns. Yet, this use of digital technology also raises substantial privacy risks, especially when one considers the possibility of “ratchet effects” that may see such tools — and their public acceptance — remain in place long after the COVID-19 pandemic crisis has ended. As we enter the age of the Internet of Bodies (IoB), whereby sensitive health data can be collected through digital devices, one must be cognizant of the fact that digital technologies could power new forms of surveillance, whereby our every heartbeat can be spied upon.
As part of an OECD Forum series, our virtual event Healthcare in the Digital Age took place on Thursday 3 December. This event has ended but you can watch the replay below.
The potential of digital technologies in health R&D is very significant too. Through their ability to rapidly analyse large volumes of research data and identify patterns, AI systems help better understand the virus and accelerate medical research on drugs and treatments. These technologies are also central to the development of personalised medicine, which promises to tailor therapies and treatments to the specific needs of each patient. AI in health does not escape, however, one of the key characteristics of machine learning algorithms: they are only as good as the quality of the data used to train them. As the saying goes, “garbage in, garbage out”. And in this respect, it is worth noting that health data do not even need to be of a poor quality to lead to poor health outcomes. Data collected from a specific population will allow an AI system to suggest health treatments that are suited to that group, but not necessarily to other people. With most health data being collected — and most algorithms being developed — in specific parts of the world, there is a risk that digital solutions could reinforce health disparities.
This ambivalent effect of digital technologies in healthcare does not only apply to AI and more advanced technologies. The significant scale-up of telemedicine in the pandemic has demonstrated the extent to which ICTs enable the remote monitoring of changes in the health status of patients. And whilst there is no doubt that COVID-19 has created a preoccupying backlog of people requiring care for other severe conditions, telemedicine may also be able to help in this respect, notably by enabling a better triage of patients. In the future, some predict that the combination of telemedicine with AI solutions might even allow for the remote diagnosing of some forms of cancer through the incorporation of real-world data into video consultations. In a similar manner, ICTs could help provide better access to healthcare to scores of people living in remote areas, or facing a shortage of health professionals. Nevertheless, the provision of such services rests upon the existence of efficient digital infrastructures and digitally savvy patient cohorts — both of which tend to be scarcer precisely among those populations who most need healthcare.
Read the latest OECD report on digital health: Empowering the health workforce: Strategies to make the most of the digital revolution
How then are we to harness the potential of digital technologies in health, whilst managing the potential risks and pitfalls? Trust is a precondition to the effective deployment of digital technologies in this sector — as the very low take-up of contract tracing apps in Europe amply evidenced. Technology itself can help in this respect thanks to privacy-preserving innovations, such as differential privacy and federated learning. Better governance of health data, compliance with international principles for trustworthy AI, strong privacy standards, and addressing unresolved questions around health data ownership are essential in this respect. At the same time, we need collaboration on a global scale to solve long-standing obstacles to data interoperability, such as the lack of agreed global standards for data exchange and terminology and the lack of even consistent patient identification in medical records in some countries.
In addition, significant investment in data management and interoperability is required. Health systems generate mountains of data, but do not routinely re-purpose these for assessing the performance and value of treatments. As the OECD puts it, the sector therefore remains ‘data rich, but information poor’. In addition, upskilling the workforce is essential. Frontline healthcare workers, for whom these technologies are meant, must be able to understand their use and have the capacity and incentives to integrate them into their workflows. The digitalisation of healthcare will fail if innovation is restricted to the R&D phase, when it must also occur throughout the implementation process.
While many sectors have long taken advantage of digital technologies to improve their services and products, healthcare still lags far behind. The OECD finds that the key barriers to building a 21st century health system are not technological. They are found in the institutions, processes and workflows forged long before the digital era. In the face of the worst health crisis in decades, it is high time to display leadership, develop responsible data stewardship, and address capacity and operational hurdles in order to realize the promises of AI and digital technologies in health.