Discussion at this event focused on the prospects, potential, state of the art and challenges associated with the extensive use of Artificial Intelligence (AI) in the healthcare sector. The growing amount of data available to the medical sector, along with the emergence of new AI systems, is making it possible to mine those data for elements capable of improving healthcare as well as accelerating research and development in the sector itself. Even more important is the fact that the combination of data and AI allows us to go beyond the concept of the human body as a “black box” – an approach, along with what is known as evidence-based medicine, that necessarily limits medical practice to observing specific output from input.
The evolution of technology and the growing availability of data now offer long-term solutions to problems regarding the sustainability and efficiency of healthcare systems around the world. A sustainability strained as much by the deep changes in healthcare demand associated with population ageing as by emerging chronic illness management requirements. It is important to note how this evolution is leading increasingly to a shift in attention from disease to the wider concept of health, founded on a holistic notion of the human body and a capacity for intervention that goes well beyond the mere treatment of a diagnosed pathology.
Although they may not yet have been absorbed into real situations, a wide variety of technologies and AI-based solutions are already available and capable of providing support for medical decision-making, the development of personalized therapies and the optimization of research and development in the area of pharmaceuticals and more. Algorithms can offer support during the various stages of treatment ranging from diagnoses, prognoses and monitoring to the most effective treatment regimes. Pathology’s use of digital twins allows for a detailed understanding of the mechanisms that underlie disease, thus facilitating the discovery of new treatments and drugs.
The foundational models underpinning new generative systems are even making it possible to get beyond one of the major limitations that have thus far delayed the full adoption of AI models: the low return on investments associated with the need to build specific models calibrated on specific data (the aim of costly interpretative activities) to perform specific functions without the possibility of the large-scale optimization of models in diverse applicational contexts.
It is equally important not to underestimate the still-existing limitations of these innovative generative models, which in some settings can lead to erroneous or entirely misleading conclusions. In that regard, human supervision has been and remains of paramount importance; as do data and their availability, quality and accessibility, not least from the standpoint of protecting investments in industry.
The management of data – the real asset on which new AI instruments are built –, respect for privacy and the formation of an adequate digital culture capable of increasing confidence in new algorithms are all essential factors in fostering the construction and development of a dynamic and competitive AI-based innovation ecosystem.
In this sense, Italy is at a slight disadvantage compared with countries that have a more robust digital culture. Of particular concern is the less than prominent role our country currently plays in the development of new regulations and the ways in which innovative instruments will go on to be implemented. Since the rules of the game are already at a more or less advanced drafting stage, the risk is that of ending up with rules of engagement already forged by actors that have taken more systematic and proactive charge of the AI issue.
The result is a paradoxical situation in which technologies, instruments, data, infrastructure and skills become increasingly available while national legislative and regulatory lacunas persist – with a European regulatory framework advancing rapidly on the impetus of more dynamic players. This, along with the difficulty of reconciling and responding to open questions on operational modes of engagement, risk Italy’s loss of competitiveness and range of action. Such an outcome would be damaging from both economic and social standpoints, and also jeopardize the long-term sustainability of the Italian healthcare system.