eHealth Tribune: Artificial Intelligence in the field of eHealth
All areas are important when it comes to improving people’s quality of life or productivity through technology, but there are some areas in which the impact of implementing ICT can be more notable, even saving lives in the medium term. eHealth has become one of Gradiant’s priorities, among other things, because of the enormous potential it has and the long way to go.
At the forefront, striving every day to include systems that facilitate and support the work of healthcare workers in their day-to-day work is the team formed, among others, by Luis Manuel Morán, Victória Cal González and Lorena González Castro.
Why is AI so important in healthcare?
LG: The use of AI is a reality in the day-to-day life of many sectors, however, it has practically no presence in the healthcare sector. It has a lot of potential that is not being exploited, partly due to the difficulty of obtaining useful data, as most of the data stored in the electronic medical record is not structured. On the other hand, regulations in this sector also make it very difficult for this type of technology to be used, although steps are gradually being taken to facilitate the arrival of these systems in clinical practice.
LM: In addition, the data available is complex and the process to extract a final product from it is lengthy. On the other hand, doctors have a lot of responsibility and technology can help them in their day-to-day work, but this technology needs to be explainable. Obviously, if they do not understand where a particular suggestion comes from, they will be reluctant to trust it. There are also many obstacles that hinder its progress.
Are these obstacles intrinsic to the sector?
LM: Of course, it’s a highly regulated sector, with a lot of safety regulations.
VC: One of our biggest challenges today is to optimise information that until recently has been stored in plain text. The way healthcare professionals record information is not easily understandable for a machine. They need structured information. Much of the existing data is stored in natural language and is not easy to structure. On the other hand, among the data that is already structured, there is a great variety of ways to organise it. This prevents the easy aggregation of information from different hospitals because standardisation has not been encouraged so far.
LG: This is normal with old data, but the fact is that today the information that is stored in a structured way in the Electronic Health Record (EHR) still minimal and there is a lot of valuable data that is in plain text. Clinical IT systems are not yet ready to store this information in a standardised way, and interoperability between systems is another pending issue. These kinds of barriers have long been overcome in other sectors, but healthcare is a very critical sector, and these processes are slower.
With all these obstacles, what is positive about eHEALTH?
LM: It has enormous potential. We currently have information on millions of people that we can use to improve healthcare, while protecting their right to privacy as much as possible. In fact, there are many systems based on artificial intelligence that are already helping doctors in different specialties to make better decisions. The knowledge of these professionals cannot be replaced in the short or medium term, but it can facilitate their daily work by suggesting diagnoses, pointing out drug interactions, minimising human error, etc. Moreover, medicine, like any other discipline, is evolving and being updated on a daily basis and technology can help our professionals to keep abreast of the latest developments and integrate them automatically into their day-to-day work. It will be a real revolution.
How do you feel about working in the health sector?
LM: For my part, there is a very important emotional component. I think it’s wonderful to help society in something as important as health. I’m delighted to work in this field, it motivates me to keep fighting against the odds.
VC: It’s something that we haven’t talked about much between us, but it’s particularly exciting for me. I think it’s really important to know that I can contribute to helping doctors do their job better. They are people who have a great responsibility and who need support.
LG: Of course it is a plus to develop your work in the health sector, and even more so when you see how much can be improved. Unlike other sectors that are mainly focused on maximising their profits, here the objectives go much further, it is about improving the health and quality of life of society.
How can AI affect the efficiency of professionals?
LM: Doctors often have quotas of six minutes per patient. In that time it is normal that they cannot examine as much as they should. Technology in general and AI tools undoubtedly speed up the work. In the end, if you can delegate the most repetitive or trivial tasks, you have more time to concentrate on the person in front of you, and you can even spend more time than stipulated on particularly complex patients. Obviously, it is not the same for someone who goes for a routine consultation, a follow-up visit or a prescription for medication, as it is for other cases that are more rare or serious.
Could this make the service cheaper?
LM: Of course. The aim is that the doctor is only where he or she is needed and that a part of his or her workload will be referred to these AI systems. Of course, the final decision will always have to be made by a doctor, but as we have already mentioned, there are currently a large number of routine tasks that can be avoided. With AI support, you get more efficiency.
What are the projects you have underway?
LG: We have recently completed the NETA V3A project, which is an assistant that helps the elderly to stay independently at home for longer by providing them with a simple voice interface that makes it easier for them to use technology. During confinement it would help them to keep in touch with their loved ones, for example through video calls, and feel more accompanied.
Also ISMAEL, which is a project focused on structuring clinical data, which as we explained is one of the main current needs in this sector.
What are we going to find in the coming years in the field of health and social care?
VC: What we believe will advance the most in the coming years is XAI, the Explainable Artificial Intelligence. It is essential in the healthcare field. There is no point in having a tool that gives very precise results if it has no credibility in the eyes of the professional who uses it. A great effort must be made to explain how Artificial Intelligence works, which often seems like magic to non-technical people. Clearly healthcare professionals are far from making decisions by simply relying on magic. They need explanations and our AI has to provide them.
LM: But we don’t just need doctors to understand and accept the advice and suggestions of a given AI. For the technicians themselves, it is also necessary to know why a certain module has gone wrong. The performance of an AI algorithm depends very much on the data that has been used to create it. By knowing in which situations the system fails, we can retrain our models and evolve the AI to make it smarter.
What could lead Artificial Intelligence to fail?
LG: The most important factor is the data that is used to train that AI. For example, if an algorithm is trained only on flu cases and is presented with a new patient with another illness, it will only be able to determine that it is not flu, but it will not be able to identify what it is. AI is like a small child, it learns from what it has seen. If the data is poor, even if the algorithm is very good, it is not going to be useful.
LM: It is because of the way Artificial Intelligence works that curious anecdotes are sometimes reported in the media. A few years ago there was a human resources system that only recommended men for management positions. It was not that the recommender was sexist, it was simply that the data available to him indicated that all the top managers at that time were men and therefore he considered it an essential characteristic for selecting new managers. The system did not really know if there was a cause-effect relationship, it was just pointing out a statistical fact. This is why it is so important to have a wide variety of data.
In my opinion, two areas of work will be key in the coming years: XAI, to dispel user mistrust; and information sharing so that models can be trained on a wider variety of data.
How has the health sector advanced in recent decades?
LM: We have come a long way in the last 30 years. For a start, we have a health card and our data is contained in an electronic medical record. We can also order our test results online. Little by little we are making progress. The health card, for example, was a very important step forward in Galicia. It is really convenient to be able to go to the pharmacy with it without the need for paper prescriptions that can be misplaced at any time.
What is Galicia’s role in Spain and Europe? For example, we have PERSIST, which is a very big project that we are piloting from here.
VC: At first one might think that Galicia, being so far away from central Europe, is technologically lagging behind, but the PERSIST experience has been very positive. Obviously there are differences between the different countries involved, both from a health and technological point of view. What is really exciting about a European project is to see that all partners have strengths, but also weaknesses that collaborative work helps to overcome. It is exciting to see Europe moving forward together.
Obviously, if we are talking about a healthcare system in which digitalisation has been boosted (for example, by promoting countless medical standards), we must take the United States as a benchmark. However, I hope that by taking advantage of initiatives such as the past H2020 or the new “Horizon Europe”, Europe will also take the lead.
LM: In my opinion, in Spain, the Basque Country and Catalonia are focal points for ICT in healthcare, but I think that at GRADIANT we are doing a great job. In the eHealth area we have a varied and very cohesive team and I like to think that with everyone’s efforts we will become a European benchmark. This is not exclusive to our team, in Gradiant we have a very collaborative philosophy. The technologies we work with are transversal to most areas of Gradiant and if we manage to be an international reference is precisely because we all row in the same direction.
