Artificial Intelligence (AI), especially generative intelligence, will not only allow us to move towards intelligent and personalized medicine, but also towards more human care. Having structured and transversal data to obtain quality results is one of the challenges for the future; another, its confidentiality and security, so that the use of that information does not represent an attack on the privacy or intimacy of the citizen. There is a long way to go and many investments to make, but the race is unstoppable.
This was one of the conclusions of a debate on the era of intelligent medicine held, in collaboration with Minsait, Indra's company, in the framework of Diálogos en La Vanguardia. The round table included the participation of Pablo Garzón, head of Sanidad en Catalunya de Minsait; Pol Pérez Sust, general ICT coordinator of the Department of Health and director of Information Systems at CatSalut; Elisenda Serra, director of the Citizenship, Innovation and User Area at CatSalut; Antoni Trilla, senior medical consultant at the Clinical Hospital and Dean of the Faculty of Medicine of the UB; and Anna Benavent, Director of Digital and Data Strategy at Parc Taulí, in Sabadell.
AI, Pablo Garzón stressed, has the “potential to transform healthcare,” but he warned that “the main enabling element” to achieve this is “access and processing of large volumes of clinical data.” The data, they all maintained, is fundamental, but until now the tools were not available to work with it properly. Serra emphasized that the citizen is the owner and responsible for his or her data and that the administrations are the ones who guard it.
There is a lot of data, the challenge is to optimize it. In the shared medical history of Catalonia there are more than 1.5 billion structured data, Pol Pérez detailed. “We have a very rich and powerful health system in information systems, but heterogeneous. Also very interconnected, we have worked a lot on data interoperability, but not knowledge,” he remarked, adding that, for now, there is syntactic interoperability, but not semantics. "And we have to go one step further, give content and context to the data so that when a professional reads the information about a patient collected by another service, they not only know, for example, the blood pressure, but also the contextual information of the data, how and when it was taken, and with all that information I can decide with better knowledge of the facts,” he stated. Catalonia, he added, is already working on semantic interoperability standards.
Pérez explained that one of the “obsessions” of the Health Department “is equity.” The Clínic or the Parc Taulí, he argued, are leading hospitals in Catalonia, but the Berga hospital also needs AI algorithms. “We don't have to reinvent the wheel, we have to take advantage of what is done in these large centers and scale and mainstream data-driven innovation with a single, centralized or federated data model,” he emphasized. “If good algorithms and good prediction tools are generated in centers that have a lot of experience, this must be extended,” agreed Antoni Trilla.
According to Garzón, Minsait has focused its efforts on interoperability solutions that “break the blockage caused by the dispersion of data, collecting heterogeneous data from different sources of information, guaranteeing the persistence, traceability and standard exchange of said information and making it available to clinicians. , managers or researchers so that they can be used safely and efficiently.”
The use of AI in health provides benefits at different levels. On the one hand, those that most directly impact the patient in the different phases of their health process, but also a very valuable tool in organizations for the management and administration of resources and productivity management, all the speakers assured. . In this sense, Trilla argued that, if AI helps simplify administrative or bureaucratic tasks, the citizen will benefit. “It will give doctors more time to talk to patients, listen to them,” added the Clínic epidemiologist. The fact that technology has more capacity and supports an infinitely greater workload than that of doctors will also favor attention. “Medicine is a mix of science and art. With much of science the machine will help us, but the art will remain for the professional.” And since AI “is not infallible,” professional healthcare supervision will always be needed. “AI can humanize medicine,” stated Trilla, endorsing the thoughts of cardiologist Eric Topol, collected in the book Deep Medicine.
During the debate, some initiatives that both the Health Department, included in the Salut/AI Program, and hospitals have, were presented. Serra, for example, explained that one of the projects underway is scheduling by reasons, a system that, depending on the reason for the consultation, directs the citizen to the type of professional who has to serve them and establishes how long it will take. has to receive. At the Clínic, we work, among other projects, on dermatology with 360-degree photos of the body to detect all pigmentations and, with the help of an algorithm, evaluate future risk.
Anna Benavent presented two initiatives from Parc Taulí. An advanced resolution unit in emergencies in which, after screening and for certain pathologies, the nurse, who is an advanced nurse, has an AI algorithm as a support tool that makes proposals for treatment, tests and diagnostic guidance.
“We do it within a clinical trial, so the patient is asked if they want to enter and signs an informed consent,” he added. The second is an AI algorithm that helps in the detection of stroke, a disease in which every second gained counts. Pérez and Benavent also highlighted the medical image repository managed by CatSalut through the IDI (Institut de Diagnòstic de la Imatge) and the Parc Tauli in the CPD (Data Processing Center) of the Generalitat and that, with its three petabytes , is one of the largest in Europe. The project, called SIMDCAT, “is an asset of Catalonia,” they stressed. The deployment of AI also requires a lot of investment.
The speakers highlighted the contribution of the European Next Generation funds, although Pérez stressed that these aid do not allow the recurrence that is needed and, in addition, generate an inflationary effect. The director of Information Systems at CatSalut advocated substantially increasing the ICT allocation. “In Spain we are between 0.8 and 1%, while in Europe the average is 5%,” he stated. Pablo Garzón, for his part, assured that there is a strong trend towards a change in the data use model by organizations, “with a high investment in data governance initiatives by the ministries, while hospitals and research centers invest in use cases.”
Legislation is another challenge for the extension of AI in health. The new European Health Data Space, Garzón argued, will allow citizens to be in control of their data and also for it to be shared securely for research and altruistic purposes. On AI, the EU will regulate all work with one law, the world's first comprehensive AI standard. EU states must select a national supervisory authority for the application of this regulation in their territory. Spain leads this European initiative with the creation of the Spanish Agency for the Supervision of Artificial Intelligence.
Regarding data confidentiality, technology is generally ahead of regulations, Benavent said. “And that is why we always find this gap. That does not mean that, with good habits and procedures, even if you go to the limit, the confidentiality of the data cannot be guaranteed,” he added. In this sense, he recalled that “zero vulnerability does not exist.” Since, as he said, “AI, especially generative, has democratized technology,” citizens also have to take responsibility. “It is not just the task of experts or technologists, it is everyone's job,” he concluded.
