In the modern healthcare system, rapidly increasing complexity of decision-making processes and their cost can be seen. This is due to the large amount of information streams (new emerging equipment, procedures and therapies, variety of drugs and generics, etc.) that translate into new treatment options, but at the same time may hinder the selection of an optimal treatment path in specific patient cases. In such a situation, the solution is to use supporting systems that make use of decision systems.
Artificial intelligence is a concept that covers a much wider phenomena than just the activity of intelligent robots. The first examples of medical activities of androids or robots partially replacing doctors (for example in surgical disciplines) obviously attract public attention, that's understandable. It moves the imagination, makes you ask a number of ethical, formal and legal questions.
Saving and extending life or improving its quality is also very attractive. And the improvement of diagnostics and therapies thanks to algorithms based on artificial intelligence tools is not just a pipe dream or a crazy theory, it is starting to happen in the world.
But such solutions should be based on very high quality databases. If we focus on this high-quality collection of the most in-depth medical data, including genomic and other -omics research, then we have a good chance to remain highly competitive on the world market of artificial intelligence solutions.
Let us remember that the data collected for many years in highly developed countries are based on a fairly standard format, referring to the most important facts related to the patient's state of health, such as time of admission to the hospital, time of discharge, type of treatment, prescriptions written, rehospitalization.
Of course, these are very important information, especially from the point of view of cost estimation in the healthcare system, but by no means does it exhaust all the possibilities of creating a high-quality database that goes even deeper.
Machine learning algorithms in healthcare facilities use data that are subject to privacy protection. Artificial intelligence is not an arbitrary creation - its operation is supervised by a person who has access to the data of sensitive patients, these data are processed, and for the system to be developed - the data must be processed again, as the dataset expands.
The origin of the data and consent to its re-use are particularly important in the case of machine learning algorithms. All data collected by the algorithm should be obtained legally, with the consent of patients, and take into account the legal regulations; among others recently adopted by the European Union (GDPR). This right applies to all EU residents, regardless of where these data are processed.
The effectiveness of artificial intelligence in individual areas of medicine is still a very open question. The output must be representative for the algorithm to be effective. If, at the very beginning, the training set will be based on low quality data and a poorly selected sample, then despite the correctly performed simulation, the results or forecast will be out of touch with reality. The biggest challenges of modern medicine, but also for the regulatory authorities, are on one hand to ensure the legality of data, but on the other - to their representativeness. Otherwise, artificial intelligence can contribute to the diagnosis or forecast burdened with errors, which is rightly pointed out by doctors and intuitively - patients.
BrainX Community-Europe was created to bring the perspective of thinking about AI as a tool that can be extremely effective in the process of supporting diagnostics.