Assistant only
Will artificial intelligence be able to completely replace doctors
Artificial intelligence in medicine today is actively used by professional and scientific communities in research and diagnosis, it participates in the development of modern medicines, learns to predict diseases and assists in their treatment. How artificial intelligence helps doctors and patients, which technologies are most in demand in In Russia and abroad, whether the use of artificial intelligence can completely replace doctors or bring a person closer to immortality, says one of the leading experts of the Russian pharmaceutical market, the head of the Eurasian division of the pharmaceutical company Bosnalek, Valentina Buchneva.
Artificial intelligence: scientist, diagnostician, assistant…
Medicine is an area of modern human life where artificial intelligence technologies are not just in demand. It is believed that the health industry is an important part of society, where AI can bring humanity the greatest benefit in the near future.
Computer vision
One of the most relevant uses of artificial intelligence in medicine is computer vision. This technology makes it possible to process the results of studies presented in the form of images – X-rays, ultrasound, CT data, images obtained through a microscope, and diagnose pathologies based on previously studied data. In particular, computer vision is used in the diagnosis of oncological diseases – several systems in the world have already been developed for the recognition of malignant neoplasms based on the analysis of the results of X-ray examinations, fluorography and CT of patients. At the same time, the AI sorts patients by urgency – those who have suspected oncology will be the first to receive advice from an oncologist. In dermatology, a platform is used to identify the most dangerous skin disease (melanoma) by analyzing photos taken with a wide-angle smartphone lens. This algorithm is able to distinguish skin cancer from non-dangerous changes with an accuracy of over 90%. ScanNav technology examines the results of ultrasound diagnostics of pregnant women to identify potential fetal pathologies. BIDMC's smart microscope based on AI technologies is able to detect fatal blood pathologies with an accuracy of up to 93%.
Neural networks
Another direction of artificial intelligence development is deep learning, when a system created on the principle of a neural network is able not only to process data according to a certain algorithm, but also to self–study, finding new patterns and offering its own conclusions. The use of AI systems based on deep learning is especially relevant when it is necessary to analyze a large amount of data in a short time.
Such systems were, in particular, tested during the coronavirus pandemic. One of the developments was created to automatically predict the risk of deterioration of the condition of COVID patients in the emergency departments of hospitals. A deep neural network "studied" X-rays of patients with positive PCR tests for COVID and offered a forecast of their condition for the next 24, 48, 72 and 96 hours. The system is configured to predict the potential use of a ventilator, referral to the intensive care unit or death of a patient. The development has shown accuracy comparable to the forecasts of experienced radiologists. With the help of such a program, according to the developers, it is possible to properly allocate limited medical resources in the conditions of a pandemic and thus save more lives.
AI also worked on the development of drugs that can block the replication of the SARS-CoV-2 virus, which significantly accelerated the search for effective remedies, which usually takes at least ten years.
Artificial intelligence, created by the type of neural network, also helps doctors predict a sharp deterioration in the health of patients. For example, AI is able to predict complications during surgical interventions. The platform, which studied the condition of more than 1,000 patients during 10,000 hours of operations, learned to predict a threatening situation 15 minutes before its onset with a probability of more than 80%. There is also an algorithm that can predict the occurrence of sepsis in patients a few hours before the manifestation of its symptoms. AI is used for the same purpose in patients with diabetes mellitus – the neural network analyzes the dynamics of the patient's glucose level and warns of the danger of hypoglycemia or the onset of diabetic coma. Neural network technologies are also actively used for DNA analysis, when the human genome system can not only calculate the growth or potential level of education of a person, but also predict serious diseases such as heart attack, stroke, cancer, bone diseases.
Speech recognition
Another area of work of artificial intelligence – speech recognition systems – are also used in modern medicine. In particular, there is a platform for early detection of signs of autism and Parkinson's disease. Another development allows you to conduct psychotherapy sessions, finding an individual approach to each patient thanks to AI algorithms. In the UK, test consultations of patients are conducted by a robot using AI technology. His task is to correctly recognize complaints and distribute the patient according to the directions of the medical institution.
Numerous cases of the use of artificial intelligence abilities – from exoskeletons capable of avoiding dangerous sections of the path of a person with limited motor capabilities without external commands, to "smart" toilet seats that can collect data and, based on them, predict the deterioration of the health of family members, notifying the family doctor about it, allow us to state for sure that medicine and machine the mind is already moving towards each other at a tremendous pace.
All major representatives of the IT industry invest in the development of AI in medicine. Russian cases
Well-known venture investor Vinod Khosla said: "Machines will replace 80% of doctors in the future and the driving force of this process will be entrepreneurs, not doctors." On the one hand, the development of artificial intelligence technologies requires significant costs – the concentration of computing power in one network, processing large volumes of primary documentation for "training" AI. But on the other hand, humanity spends enormous funds on maintaining health – the market for medical services in the world is approaching a trillion dollars in value, pharmaceutical services exceed 1.5 trillion, the market for wearable mobile devices for health monitoring in 2020 amounted to $23.1 billion, and by 2027 it may exceed the figure of $ 250 billion.
In the world
Therefore, today the world's largest corporations are creating their own software products and buying successful startups in the field of digital medicine. In 2020, the volume of investments in this market was estimated at $ 4.5 billion with an annual growth forecast of 40%. For example, IBM has developed the Watson Health platform, which is actively used in radiology. Google has offered British doctors its development in the field of ophthalmology DeepMind Health. Smartphones running on the Android system use the Google Fit app to collect and analyze data about the owner's activity. Microsoft Corp announced its development of AI for Health, under which it will invest $40 million in artificial intelligence (AI) technologies for the healthcare sector for five years, Apple has been improving its Apple Health application since 2014, expanding the ways to collect and analyze information about the health of the owner of a mobile device. Transactions are also conducted in the opposite direction – large medical companies cooperate with IT concerns (like Novartis and IBM) or buy IT developers (like Roche, which bought the developer of medical technologies Flatron Health). In general, there are more than 800 companies in the world leading the development of AI systems applied to medicine.
In Russia
Today, according to the annual almanac "Artificial Intelligence", 21% of all investments directed in Russia in artificial intelligence development technologies are in the development of health products. For example, the Russian TeleMD platform diagnoses oncological diseases, the domestic development of Cels also works with the results of mammographic and fluorographic examinations, CT scans of the lungs and histological studies to detect oncological diseases. The Center of NTI "Artificial Intelligence" is testing a prototype of a robot arm for patients with reduced capabilities and loss of hands. The patient will be able to control the robotic arm attached to the body with the help of a neural headset with an eye tracker. The world-famous Da Vinci robot surgeon with artificial intelligence is successfully working in one of the Moscow hospitals.
Today, Russia is actively testing programs to support medical prescriptions that reduce the likelihood of medical errors. In Russia, AI is also used to decode X-ray images as part of the "Second Opinion AI" service. The success of the pilot project to introduce the capabilities of artificial intelligence into the work of medical institutions in the city of Muravlenko on Yamal inspired the introduction of similar technologies in other regions of Russia.
Another interesting domestic project with the use of artificial intelligence technologies is in the R&D stage. Health Heuristics is planned to be implemented as a platform that combines data from survey results, information from electronic medical records, measurements obtained through wearable devices and other sources about human health and lifestyle. At the time of launch, the program will have to analyze the data of 120,000 respondents. IT industry experts believe that the domestic AI market could develop at a faster pace with additional support from the state. Great hopes are pinned on the National AI Development Strategy until 2030, approved by presidential decree in October 2019. According to this document, in the period 2021-2024, it is planned to spend more than 31 billion budget funds on the development of AI technologies to improve the quality of life.
Can AI replace doctors in the future?
Neither doctors nor programmers undertake to assert that the capabilities of artificial intelligence will be able to provide a complete replacement of a doctor with computer technology. At least for the foreseeable future. To date, one of the limitations in expanding the use of AI capabilities, IT market experts call the imperfection of the legislative framework, in particular, the methods of identifying the patient's identity and establishing his consent to the collection of personal health data, the system for evaluating AI test data in medical practice. One of the steps in solving this problem has already been taken – a draft of the first national GOST R has been developed, which will be used in evaluating data from clinical trials of medical AI systems in Russia. Six more standards regulating the development of AI in medicine are being prepared for release. An obstacle to expanding the capabilities of AI in medicine today is also the high time and financial costs of entering primary data into systems for analysis, as well as the need for direct participation of physicians in developments, which is often difficult to organize due to the high employment of qualified practitioners. The introduction of AI into medical practice around the world is also hindered by the insufficient number of trained personnel who will serve systems equipped with AI technologies.
The insurmountable limitations in the use of AI today are due to the fact that diagnosis and choice of therapy is always a creative, nonlinear process that requires such a broad and multidimensional set of skills from a medical specialist, which is difficult to achieve in machine intelligence systems. In addition to these obstacles, there is one more psychological one: personal contact between the patient and the doctor, the patient's trust in the physician in many cases have an additional therapeutic effect that cannot be discounted for at least several more generations.
Conclusion
Scientists involved in medical forecasting assume that, based on the current pace of development, after 2045, humanity will annually add a year to the average life expectancy. Improving the genotype with the removal of "dangerous" mutations, replacing worn–out body structures, reducing the likelihood of medical errors, reducing the time to search for new drugs and vaccines, reducing the routine burden on doctors and freeing up their time for an individual approach to the patient - these are just an incomplete list of opportunities that already improve the quality of human life and prove the need for a broad applications of AI in medicine.
Portal "Eternal youth" http://vechnayamolodost.ru