Artificial Intelligence (AI) is used in practically all areas of our lives. No one can avoid hearing about it and the challenges it faces. But are we really aware of all the possibilities generated by AI? In the field of medicine, experts affirm that it is “a revolution that is here to stay” with “infinite possibilities”.
AI has the potential to bring immense benefits to a variety of fields, from accelerating medical discovery to Drugs And treatmentImproving diagnostic accuracy, personalizing or optimizing treatments data management Doctor. To know what we are talking about, we have professionals to admire who lead us on a path of innovation that is difficult to understand.
«Artificial Intelligence offers two complementary approaches that relate to two types ofKnowledgeDifferent but unified for the first time in history: predictive and explanatory knowledge», explains Oscar Pastor, researcher at the Valencian University Institute for Research in Artificial Intelligence (wren in English) from UPV.
«The prediction dimension is magic of mathematics», the priest announced. With predictive technology, science is able to train algorithms to predict diseases from data with analytical techniques. machine learning, «One of the projects we have in Wren is to predict the appearance of colorectal cancerHow is this achieved? utilizing the vastness of data, With a sample of patients with this type of cancer and other healthy patients, they analyze and compare the data and the ‘magic of mathematics’ is what makes it possible for any new patient to predict whether they will have the disease or not. .
“The amount of data that can be stored on each individual is enormous,” says Jose M. Sempere, WREN’s deputy scientific director. As a result, the process ofpersonalized medicine»Since it allows profiling patients, understanding each individual’s mutation and optimizing treatment. “This is very clear in oncology, it is becoming more and more personalized in this area and in the short term we are going to see very promising results.”
“AI algorithms can analyze large sets of medical data, such as images magnetic resonancetomography, This may help improve identification and treatment success rates and reduce potential “medical errors” in decision making.
Hospitals worried that such a large amount of data breach confidentiality From patients. “The aspect of data privacy in health must be very clear to avoid problems,” says María José Galindo, head of the infectious diseases unit at the University Clinical Hospital of Valencia. “We need the data to be validated by professionals and, in addition, patients can clearly accept the conditions presented by AI.”
According to Dr. Galindo, AI in consultations is “low in reach” at this time but shares experiences that have occurred in other hospitals. «The Hospital Clinics of Barcelona, together with the Barcelona Supercomputing Centre, developed an algorithm to predict the evolution of patients admitted for COVID-19. In this way, it was possible to define what werecovid disease patternAnd what the patient was going to develop and what kind of treatment should be done individually, because there is no single Covid, but it can affect for example lung inflammation or vascular problems.
For his part, researcher Sempere points out that AI techniques are important for drug design. «With Artificial Intelligence, the testing and testing of drugs is accelerated, the type of mutation we want to reduce is outlined and we try to establish combinations of possible drugs and see their effects, we Also does screening with AI before proceeding clinical trials, Similarly, Nuria Oliver argues that “the design of clinical trials can also be improved by identifying more effective inclusion or exclusion criteria, predicting treatment response, and optimizing patient recruitment.”
“It will no longer be a medicine for everyone, but rather for each patient to adapt the treatment that is best suited to their type of pathology.” Dr. Galindo explains that this model individual attention This already happens in HIV counseling and other infections or chronic diseases and AI can be a good ally to move forward in this sense if it is used with supporting informationProfessional”. “Twenty years ago it was unimaginable to design molecules in laboratories, certainly in the future they can be created thanks to AI” vaccines In a simple way.
There is a limitation to this predictive dimension: “The algorithm is able to predict, detect, and treat, but it will not be able to explain why you have that disease.” AI uses the interpretability dimension to figure out where the problem comes from. “The fact that the explanatory part specifies what problem you have, we understand the meaning of code of life», highlights Professor Oscar Pastor.
If you understand that law of life and we realize what happens to us, we can see it as software errors, and with the correction of those errors genetic engineering “We will also have the possibility of curing diseases in a way that has never been possible in the history of humanity, that is, treating the problem at its source,” says Pastor. Artificial intelligence makes possible what previously was not possible with traditional technologies. What was “impossible” is now “achievable”.
However, this fact causes other Challenges to level Moral And Economic Because, according to researcher Pastor, this type of operation, which would involve manipulation of the genome, “would be extremely expensive” and would not be available to everyone. “Public systems will not be able to absorb them in a generalized way, so there is a risk of creating inequalities.” We have to deal with these types of challenges from the very beginning.”
To the challenges of privacy, data integrity and economics, the co-founder of Alice Alicante adds “a lack of transparency» Because of “the difficulty in understanding how these complex neural networks, which can have hundreds of billions of parameters, work”. In the medical context “it is necessary to develop AI algorithms that are explainable, so that humans can understand the reasons behind the recommendations or diagnoses provided by these systems.” Another important challenge is to “get it right”. Discrimination And this algorithmic bias, José M. Sempere, WREN’s deputy scientific director, exemplifies this situation with age bias, saying that “age bias in COVID may have excluded older people who were able to recover with adequate care. ”
Despite the challenges of AI, experts’ assessment is positive. In hospital This represents a “relief” because of the speed of some administrative processes that previously were “more tedious,” says Galindo. “It used to be very difficult to get information, now all these processes are expedited and we can spend more time with patients and optimize their care.” If everything crystallized well, it could make our lives a lot easier.”