High-level dialogue: Ensuring trustworthy healthcare in an AI world (ITU)

Event report

The session highlighted the importance of artificial intelligence (AI) to improve healthcare systems; regarding the provision of more affordable services, distribution of resources in underdeveloped communities, and responsive solutions of diagnosis and treatment decisions. However, these improvements should be followed by trustworthy and peoplecentric approach.

Concerning the use of AI technology in healthcare, Mr Houlin Zhao (Secretary-general, ITU) stressed that it was pivotal in addressing the COVID-19 pandemic. AI has been mainly implemented in tracking apps and diagnosis technologies. Nevertheless, Zhao indicated the importance of addressing issues related to security and trust in these systems before their widespread use by healthcare professionals. In addition, technology and skill gaps might be bridged worldwide in order to enable a more inclusive approach to AI in health systems. The developing world still lacks basic digital skills, infrastructure, and data required for the successful implementation of technology into diagnostics and more affordable medical treatments. To this end, health professionals, ICT teams, and government might work in collaboration.

Dr Naomi Lee (Senior Executive Editor, The Lancet) asked the panellists about the importance of collaboration between different society sectors to improve safeness and inclusiveness as the main guiding lines for AI and healthcare.

H.E. Mr Omar Bin Sultan AI Olama (Minister of State for artificial intelligence, digital economy and remote work applications, United Arab Emirates) noted that trust, safeness, but also inclusiveness might guide the usage of AI in healthcare. He underlined that regarding inclusiveness, a global approach to the development of technologies might be necessary, considering that an increasing number of countries are importers of AI technologies. In reality, very few countries are currently exporters of AI technology. Monopolies were created across the world. This scenario does not help make AI available to everyone. Furthermore, the COVID-19 showed that health problems concern all countries together. An isolated approach to health problems responses is not suitable in a globalised world. In the UAE, policymakers are working to fill the gaps related to AI. The first gap concerns the lack of understanding. It is not possible to regulate such technologies without full understanding. With this purpose, the country has signed a partnership with the Oxford University to build a holistic understanding of AI. The second gap regards the lack of data. AI cannot be fully used in the country because there is no sufficient data. So far, they have planned to invest in technology to fill this second gap.

H.E. Ms Justyna Orlowska (Prime minister’s high representative for government technology, Poland) approached the importance of small and medium enterprises (SMEs) in the sector. SMEs have more flexibility which is essential for innovation. In her view, the public sector might heavily invest in SMEs and start-ups because they are responsible for most developments in the area. In this sense, the collaboration between SMEs and the public sector is fundamental to advance AI in healthcare, because the latter has way more procurement budget. Poland has implemented the program GovTech to facilitate public investments.

Dr Ing Thomas Wiegand (Executive director, Fraunhofer HHI, Germany) has worked to fill the gaps between different communities. He realised that ICT, epidemiologists, health professionals, and the economist communities work in an isolated fashion. There is a need to create a framework that empowers those communities to work together. Only with collaboration, algorithms may be largely trained to test health data.

Dr Petra Wilson (CEO, Health Connect Partner) underlined that the use of AI in healthcare has three main gaps. The first gap relates to data science. So far, there is not enough wealth data to address and efficiently prevent issues such as pandemics. The second gap relates to skills. Clinicians might be supported to use AI. They are not fully confident yet. Finally, citizens shall be properly informed on how these technologies work, otherwise they will not trust them and will not consent to provide their health data. Regarding this last aspect, it is not enough to inform citizens about algorithms in technical terms, because citizens, in general, will not grasp it. The transparency aspect might relate to informing society how private companies use health data, for how long, and for which purposes. If transparency is achieved, citizens will be more prone to accept the benefits of AI in healthcare. She also stressed that the concept of ownership is not completely suitable to health data. The usage of health records in AI systems only exists because someone properly trained translated these records into something understandable that allows technologies to predict and diagnose. Health data and, particularly, medical records should be more properly addressed in terms of good stewardship. Good governance of medical records is a condition to implement AI in health care.

Panellists agreed that transparency, safeness, trustworthiness, and inclusiveness might be the guiding values to AI in the health sector.