A word with:
“AI applied to healthcare in critical situations
that can compromise patient’s safety is
in its early stages”
A Word with:
“AI applied to many fields, but especially to healthcare in critical situations that may jeopardize patient s”afety is in its infancy.”
Horizon scanning or technological foresight in healthcare is not something that is new today; it is a process that was implemented many years ago in the world and in which the Basque Country joined by forming part of an international organization such as EuroScan /i-HTS since 1998. Knowing what the technological trends are and anticipating them with information is a competitive advantage to be able to respond in a more informed way to the technological challenges that are foreseen and to be able to establish priorities on the basis of local needs. It also improves the relationship with innovation suppliers and innovators, since the dissymmetric knowledge gap is smaller and it is possible to talk on the same level of information, which results in more satisfactory agreements on both sides. It also allows proactive dialogue exercises to be carried out that can result in the improvement and adaptability of technologies under development in processes known as early advice and dialogue.
First of all, it is necessary to have an understanding of the needs of each of the healthcare systems (needs are context-dependent) in order to harmonize what is developed with the gaps in the healthcare systems. In this sense, the role of patients, professionals and managers is fundamental. Many promising developments have been set aside because they did not respond to needs and because they did not involve stakeholders in their conception, design and production. In the case of the evaluation processes carried out by our Osteba evaluation service and in our system, the participation of patients, professionals and managers is carried out systematically, which ensures that the process has more guarantees of success.
It was born as a result of the organizational transformation of the Basque health system itself and to respond to the new reality that was imposed with a transition to an integrated health system, working in a network and putting innovation and health research at the forefront of the agenda of the Basque health system. The healthcare system was changing, and research and innovation could not be left out of these changes. The objectives currently set are related to the consolidation of the system and the new reality with 4 research institutes established, a more consolidated national and international presence and new challenges such as new advanced therapies, personalization, additive manufacturing, artificial intelligence… These are trends that we have identified in our horizon scanning system and that should be considered for implementation in our system in the best possible way and based on our needs and priorities.
One of them is trust, especially in the case of developments based on Artificial Intelligence (AI). There is already talk of AI devices for decision making of three types: informative, prescriptive and autonomous. Certainly, we are at a point where decision aids are well accepted, especially when there is a certainty that practices that do not add value without risk to safety will be eliminated. This is being implemented in image screening where those that clearly have no trace of malignancy or pathology with 100% true negatives are certainly eliminated. What is less clear is when we move to the diagnostic rather than the screening side or when we endow the system with autonomy. Given the new developments, some derived from the combination of data and their analysis by neural networks that learn continuously, the frontier here is in two planes: a) in the curation of the initial data that feed the networks and b) in the conviction that the models do not count to increase their validity with confounding variables that maximize one branch over another and induce error. In these cases, the explanations of AI-based models are not as simple as those of more classical statistical models. AI applied to many fields, but especially to healthcare in critical situations that may jeopardize patient safety is in its infancy. However, informative or prescriptive aids linked to safe filtering processes do represent a current competitive advantage. It is even being considered when alternatives may have higher risks. Here another fundamental role will be played by legislation and legislators, since the ultimate responsibility lies with the producer and not with the user, except in flagrant cases of negligence in use. Other developments will affect devices such as possible amendments to EU regulations 745 and 746 concerning medical devices and diagnostic tests. In addition to the conception that diagnostic devices cannot be unbundled from differential treatment or management of patients when we are talking about end results. The role of additive manufacturing in healthcare (allowing greater personalization) and minimally invasive therapies, including robotics, should be considered at a closer level because they are already beginning to be standard practice in some cases.