Medical informatics is a huge branch of an interdisciplinary field at the intersection of computer science and medicine, which, interestingly, is very large both from the point of view of computer science specialists and from the point of view of medical specialists. The most general definition that can be given is the processing of data that results from medical activities and the support of any healthcare processes. At the same time, healthcare itself is a huge field, and it is necessary to understand what place medical informatics occupies in it. Medical informatics and its development are inextricably linked with evidence-based medicine. In principle, computer science as such was able to develop together with medicine when we had the methodology of evidence-based medicine. From the history of evidence-based medicine, we can see that it is inextricably linked to data processing and medical statistics. Evidence–based medicine is a medicine that produces any effects on a patient only when there are preliminary grounds to believe that they will be beneficial, and these grounds are obtained by a scientific method. The scientific method uses statistics to make verifiable and repeatable judgments. Hence, in medicine, we achieve desacralization, repeatability, and statistically investigate this. The process of statistical research consists primarily in controlled randomized trials, that is, in proving on serious samples that one effect is better or at least not worse than another. The corresponding methodology is called a superiority study and a non-inferiority study. When we realize that we need statistics, we have to work with the raw data and with the processed data. This is what underlies medical informatics. The question immediately arises: where does the data come from that tells us that something is better and something is worse? We need an objective diagnosis, which is a separate subdomain. After diagnosis, we receive a huge amount of raw data that needs to be converted into other data, linked to the patient, medical history taken, and so on. And this is another subdistrict. As a result, today it is considered that medical informatics is divided into clinical informatics, that is, the real support of computer science for the treatment process itself in the clinic, and what is called nursing informatics in English – this term translates poorly into Russian, since it is difficult to make an adjective out of the word “nurse”, so they usually say “optimization of processes in the clinic”. medicine”. That is, it is, in fact, standard office work and patient lifecycle management, operation lifecycle management, and so on. There is a very interesting additional field – biology, more precisely, bioinformatics and cheminformatics, which gives us the opportunity to make medicines and understand what is happening to the human body. Currently, bioinformatics and cheminformatics distinguish a particularly important subsection, which is called differently, but is related to genetics. In turn, this entails personalized medicine and another branch of medical informatics. The basis of all this in mathematics is medical statistics. Thus, we have a lot of levels of abstractions – from mathematical disciplines (statistics, graph theory, and so on) before applying this to large systems based on, say, systems engineering. And if we are currently engaged in evidence-based medicine, then we are automatically engaged in medical informatics, and vice versa. Without evidence-based medicine, everything in medicine loses its meaning. They also say that in biology everything loses its meaning without evolution – and vice versa. So, we have slightly classified the areas, and now we need to understand how they are supported by modern information technologies. The Internet allows clinical centers to communicate well, conduct randomized trials around the world, test medicines that are made in one part of the world in another, and so on. This also makes it possible, with well-established standards and special treatment procedures, to do what is now called telemedicine. This is a very serious technological basis. In Russia, since the beginning of 2018, a law has been in force that prescribes telemedicine, although it also has its drawbacks. Currently, the legislation is very far behind the opportunities that medical informatics provides to medical professionals and, first of all, to clinicians. The process of harmonization of this legislation is underway all over the world. But legislation is lagging behind for a reason, primarily because when we talk about computer science, we’re talking about data processing. The data must be somehow received, stored, transmitted, then processed and used. These are very complex processes in which the principle of medicine “do no harm” must be used. The source data is usually noisy. The quality of this data is often very poor when new information technologies are introduced, and it takes some time before it gets better. And generally speaking, data in medicine can be completely different. There is raw data that accumulates with any impact on the patient and any diagnosis. There is data that is knowledge about what is good and what is bad and how to do what, and this, on the contrary, is the upper level. There are several more layers between these levels, that is, the knowledge level and the raw data. The first level, following the raw data, is data that is handled by basic medical information systems and that does not require a medical professional as such. These are personal data, data on visits to clinical institutions, data on the facts of operations, examinations, and so on. We have been able to process this data well for a long time. But then there is data about what happens to the patient, and this is diagnostic data. Nowadays, not everything is objective diagnostic data, there is a lot of subjectivity. The data of objective diagnostics first appear in a raw multimedia form, for which, fortunately, there are already good standards. And we get the opportunity to create a special class of systems, for example, PACK systems, which accumulate this data. Then the data needs to be processed. This is now a very good direction of development, in which amazing successes have been achieved. For example, artificial intelligence methods have built systems that can make better diagnoses based on diagnostic images than the average doctor in this field. Moreover, in some subdistricts, such as oncology, these systems diagnose better than the consultations of the best doctors. It is clear that these systems are not allowed to work at full capacity. And doctors are still trying to test algebra with just something that is an art. According to my expectations, it will be only by the 2030s, both after the harmonization of legislation and after debugging the relevant systems, that this will become part of the usual medical practice. But since 2017, for example, programs have been operating in the United States that include personalized medical services in insurance and are fully developing. It should be said that after we process the primary patient data and diagnostic data, we need to aggregate them and draw conclusions based on them at the regional, national and global levels. The World Health Organization supports several standards and a huge number of tools for working with such aggregated data. It is especially interesting that we have beautiful ontologies, that is, a formalized description of knowledge at this level. It should be noted, of course, the SNOMED ontology, which is now the basic one for describing everything in medicine, and ontologies related to disease classifiers, such as, for example, ICD or ICD, the international classification of diseases and health–related problems. As soon as there is a coherent system of knowledge supported by the whole world about what is now proven in medicine, then everything else is slowly being integrated into it. And based on these high-level ontologies, doctors produce more and more data that is initially tied to the meaning of what is really happening to the patient. Now there is a lot of what is called open data in medicine. This, in turn, gave an additional impetus to the development of medical informatics and allows us to independently assess how well certain methods work. This includes, for example, automatic diagnostic methods, diagnostic methods, medical decision-making methods, and so on. By the way, the software systems themselves, which are called CDSS systems – support for medical decision–making – have become the main class of systems that dramatically increase the effectiveness of clinical activities. And now it is often said that they are the only ones developing within the framework of clinical informatics. Although this is incorrect, they are simply the most visible. The classic example now is a family of products called IBM Watson, the famous IBM company. And there’s Watson Oncology, Watson TTT, which are interesting because it’s a whole family, and most of the input data for them is text. When we learned how to work with converting natural language text into some kind of more or less formalized knowledge, we got a unique opportunity to go through the entire history of medicine and turn it into something formal that a computer can use. This is happening right before our eyes, but it started sometime in the 2010s and is projected to lead to the emergence of consolidated ontologies by 2030. As a result, medicine ceases to be the medicine of a doctor, but becomes the medicine of devices. Many people don’t like it. Hence the surge of interest in all kinds of alternative medicine. But how long has it been noticed what is called alternative medicine, which has shown its effectiveness? Medicine. And as soon as alternative medicine proves that it works, it becomes medicine. And the one that doesn’t prove it remains a sect. And medical informatics, of course, cannot work with it, although it is actively trying in the sense of imitating activity. This is especially scary in Russia, because there is a huge market for obscure devices and obscure software that mimic real medical informatics software and equipment. It’s really scary. But, fortunately, it has no effect on the development of the industry itself. Moreover, the successes shown, for example, by modern personalized medicine in the treatment of oncological diseases, as well as the appearance of the first clinical studies and evidence of changes, for example, in the genome of an adult, show that this trend is not just a trend for the distant future, but a trend that will be applied everywhere in a few years. Separately, we note the support of such areas as translational medicine, medicine related to embryonic development, and preventive measures. The development of medical informatics in all these fields has led to an explosion of new technologies. Hence the cheap means of genotyping, very good means of preliminary control, and a series of studies on the embryo genome and embryonic development that began in 2017. What can be presented now as the undoubted successes that are available to an ordinary person? First of all, this is what has entered our smartphones, smartwatches, and so on – this is wearable medical equipment. It appeared largely against the will of official bodies and is being developed by corporations without waiting for the harmonization of legislation. But since such equipment shows high efficiency, it is developing faster and faster. Hence the support for people with chronic diseases, primarily diabetics: each such patient can download several programs that predict what can happen to them and what is best to do to prevent this from happening. Another area is monitoring. It has become especially popular among the elderly, among people who have switched to outpatient treatment. In addition, there is diagnostic equipment that can be used by the patient himself. This is unique, it has never happened before: previously, the provider of any medical service was a doctor. That’s not how it’s going to be right now. Finally, there are systems for continuous transmission of patient information to the clinical center and tracking with predictive analytics. When we realize that we can use artificial intelligence to predict what is possible and highly likely to happen to a patient, we finally get the opportunity to tell the patient about it, without waiting for any events related to his arrival at the clinical center or with some kind of examination. And this is a huge breakthrough. It is associated with the improvement of the situation with certain orphan diseases and the need to spend a lot of time in the clinical center. It often makes no sense now, although it remains a serious problem in Russia. Beds in hospitals are often in short supply, not because there are people who really need this help, but because they simply do not want to let them go to be treated at home, although there are all the conditions for this. There are services that offer patients electronic records management, and without taking into account any specific clinical centers, for conducting all diagnostics, accumulating it in standard formats and using this entire digital footprint of human health to improve subsequent medical services. In the coming decades, that is, in the 2020-2030s, we will see a restructuring of the standards of medical care based on integrated healthcare support systems. Author: Alexey Neznanov – Candidate of Technical Sciences, Associate Professor at the HSE Faculty of Computer Science.