Translated by
2019/04/23 23:36:33

Artificial intelligence in medicine

Artificial intelligence technologies in a root change a world health care system, allowing to process radically the system of medical diagnostics, development of new medicines and also in general to increase service quality of health care at simultaneous expense reduction for medical clinics.


Main article: Artificial intelligence (AI, Artificial intelligence, AI)

  • At the level of design: forecasting of diseases, identification of groups of patients with high risk of diseases, the organization of preventive measures.
  • At the production rate: automation and process optimization in hospitals, automation and increase in accuracy of diagnostics.
  • At the level of promotion: management of pricing, risk reduction for patients.
  • At the level of providing service: adaptation of therapy and composition of drugs for each certain patient, use of virtual assistants for creation of a route of the patient in clinic or hospital.

Artificial intelligence in radiology

Main article: Artificial intelligence in radiology

The artificial intelligence is actively applied in researches of development of techniques of diagnosis of cancer. In more detail in article:


CB Insights: In 2021 the market of medical AI technologies will reach $6.6 billion

For the beginning of 2019, according to data of CB Insights analytical company, since 2013 the international technology startups developing artificial intelligence technologies managed to attract $4.3 billion within 576 transactions. Besides, experts claim that within the next three years the market of medical AI technologies will reach a point of $6.6 billion, increasing every year by 40%.

IBM and AstraZeneca created the neuronet foretelling heart attack

At the beginning of March, 2019 the companies IBM also AstraZeneca provided neuronet which can predict heart attack. Results of work of new technology are described in the published article "Clustering on the basis of Results of Patients with a Sharp Coronary Syndrome when Using Multitask Neural Network".

Specialists of IBM and pharmacological company AstraZeneca developed a framework of machine learning which notices the coming signs of a sharp coronary syndrome
Specialists of IBM and pharmacological company AstraZeneca developed a framework of machine learning which notices the coming signs of a sharp coronary syndrome

The team of researchers collected data on age, sex, the anamnesis of life and a disease, addictions and also results of laboratory researches, information on the carried-out treatment and nearly 40 other indicators among 26,986 adult hospitalized patients in 38 city and rural hospitals of China. All data were loaded into a neuronet which had to learn whether the serious unfavorable warm event (MACE) was noted at the patient in the past and also whether he received antitrombotsitarny medicines, beta-blockers and statins – the medicines reducing manifestations of coronary insufficiency and preventing a myocardial infarction and a stroke.

Further authors of article carried out a clustering by method of k-averages for distribution of patients by seven groups on the basis of the data obtained by a neuronet. As a result it turned out that in the first cluster which supported patients with frequent cardiovascular events as a heart attack and a stroke, but as low occurrence of coronary heart disease, the main predictor of the following heart attack served presence of diabetes while in other cluster which included patients with the heavy course of cardiovascular pathology without the previous heart attack advanced age and the increased systolic arterial blood pressure were the main predictors.

Researchers warn that though the clustering matters for the forecast of a disease, not clearly whether these data in clinical practice can effectively be used. Nevertheless, their work shows that cluster analysis on the basis of artificial intelligence is perspective approach for classification of patients with a myocardial infarction. Future researches will focus on determination of "cluster and specific" interventions at which the efficiency is considered.[1] of the previous treatment.


the Size of the market of AI technologies in health care was $1.4 billion - Zion Market Research

In 2018 the amount of global market of AI technologies for health care reached $1.4 billion, counted in Zion Market Research analytical company. It is expected that by 2025 the indicator will grow to $17.8 billion, and expenses on such solutions will increase approximately by 43.8% annually.

Most of all spend for medical artificial intelligence (machine learning, context-dependent calculations, natural languag processing, computer vision, speech recognition) in North America. Leadership is caused by the fact that this region is represented by such technology giants as Microsoft, IBM, Google, Nvidia, Amazon, Intel, General Electric and Xilinx. Besides, in North America transactions on merges and absorption, large partnership and start of important products are frequent.

In 2018 the amount of global market of AI technologies for health care reached $1.4 billion, counted in Zion Market Research analytical company
In 2018 the amount of global market of AI technologies for health care reached $1.4 billion, counted in Zion Market Research analytical company

In Europe by 2019 the market of the artificial intelligence used in the medical purposes can be considered arising. In 2016 its volume was measured by $320 million, to the 2019th it will be $1.61 billion. At the same time 21% of medical institutions in Europe plan purchases of AI tools, the data of the European community of electronic health care published in April, 2019 demonstrate.

One of main the catalyst of demand for AI products in medicine is deficit of doctors. According to World Health Organization, by 2019 57 countries lack about 2.3 million nurses and doctors. The factor constraining development of this market, experts call absence of qualified specialists who could follow the guidelines in the field of AI.[2]

Carry the following companies to number of the solutions AI largest producers of analytics:

The artificial intelligence increasing success of ECO by 20% is provided

At the end of December, 2018 experts from the Cornish university in the USA and Imperial college in London showed results of the research according to which the efficiency of ECO can be increased for 10-20% if to use artificial intelligence for quality evaluation of embryos. Read more here.

The beginning of installation in China of 4 thousand boxes with the AI doctors making diagnoses in minutes

At the end of November, 2018 the largest online provider of medical services in Kitayeping of An Healthcare and Technology told that he is going to set several thousands of AI clinics of the size of a public callbox and to extend them through the whole country in three years. The first such points of delivery of health care already earned. Read more here.

As the artificial intelligence will develop in medicine in 2019

In November, 2018 the DataArt company specializing in services of consulting and development of IT solutions submitted the forecast how the artificial intelligence will develop in medicine in 2019.

According to experts, the artificial intelligence will remain an object of interest of both investors, and  health workers. AI algorithms still develop, become quicker and  more precisely. At the same time only a few pharmaceutical companies integrated solutions  on the basis of artificial intelligence technologies in  the processes.  In most cases, such solutions are used only in  pilot projects and   did not receive due deployment yet. The health care in 2019 waits for progressive and  non-standard views which will show how  fully to use all opportunities of AI.

The artificial intelligence will remain an object of interest of both investors, and health workers
The artificial intelligence will remain an object of interest of both investors, and health workers

Thanks to artificial intelligence "smart" telemedicine services will make qualitative medicine more available for a wide range of people and will help them to prevent development of chronic diseases thanks to timely consultations with the doctor.

Analysts are sure that the algorithms connected with collecting, processing and data storage in 2019 will be of great interest and the importance to the health care industry. Thanks to sensors of new generation continuous monitoring vital important indicators of health of patients already became a reality.

Modern physical diagnostic examination provides much more parts, than  30 years ago.  It includes data from  different sources  — from  family history to  concentration of protein in  a blood sample. 

Data from mobile devices create a dense data stream which needs to be processed and saved, and the 2019th should become year when progress in this direction amplifies.[3]

Japan builds AI hospitals for solution of the problem of shortage of doctors

In August, 2018 it became known that the government of Japan, with assistance of business and scientific community, begins construction in the country of hospitals in which physicians will be come to the rescue by artificial intelligence. Due to AI technologies it is supposed to cope with the shortage of doctors in Japan, to unload personnel and to cut down medical expenses. Read more here.

The first recommendations about use of AI in the field of health care are offered

On June 18, 2018 the American medical association (AMA) offered the first-ever recommendations for use of artificial intelligence in the field of health care. In the statement which the representative of AMA announced at an annual conference in Chicago the main directions of further development of AI in this industry are specified.

According to this statement, AMA she intends to implement practices in the field of artificial intelligence and other priority areas for improvement of results of treatment and for professional satisfaction of doctors. AMA is going to use the significant provision in the industry for involvement of producers, determination of priorities in development of AI and also the problem solving connected with validation and implementation of new techniques. Besides, AMA intends to develop the training plan of specialists and the report of information to patients on restrictions and opportunities which are characteristic of this category of analytical tools.

The American medical association (AMA) offered the first-ever recommendations for use of artificial intelligence in the field of health care
The American medical association (AMA) offered the first-ever recommendations for use of artificial intelligence in the field of health care

AMA supports integration of carefully thought over, high-quality and clinically approved techniques of use of AI and also requires proper professional and government supervision of their safe, effective and legal use. Analytical technologies on the basis of AI, AMA considers, should be available to check and identification of systematic inaccuracies at all development stages, to conform to the leading standards of reproducibility and also to protect the interests of individuals and confidentiality of personal information.

AMA considers that needs of users should be the focus of attention, and use of the AI system should be checked on representative selection within clinical trial.

The combination of the AI methods and irreplaceable experience of the clinical physician will undoubtedly improve the result of therapy, - the board member of AMA Jess M. Erenfeld considers (Jesse M. Ehrenfeld). – However we should participate directly in the solution of all problems arising at design, assessment and implementation of these techniques, every year the field of their application becomes wider.[4]

AI taught to predict falling of arterial blood pressure during transaction

In June, 2018 the results received by group of researchers which developed the prediction algorithm of potential hypotonia or abnormal falling of arterial blood pressure during transaction were published in the Anesthesiology magazine.

For creation of an algorithm researchers used machine learning technologythe artificial intelligence analyzed data of 1334 patients during which transaction registration of arterial blood pressure – in total was made 545,959 minutes. On the basis of these data the prediction algorithm of hypotonia was prepared during transaction.

The AI algorithm for prevention of the complications connected with hypotonia such as postoperative myocardial infarction or sharp renal failure is created.
The AI algorithm for prevention of the complications connected with hypotonia such as postoperative myocardial infarction or sharp renal failure is created.

Having approved this algorithm, researchers carried out its inspection on the second data set including indicators of arterial blood pressure of 204 patients with a general duration of 33,236 minutes. 1923 episodes of hypotonia entered these records. The algorithm precisely predicted sudden falling of arterial blood pressure in 15 minutes prior to its emergence in 84% of cases, in 10 minutes prior to its emergence - in 84% of cases and in five minutes prior to its emergence - in 87% of cases.

Researchers assume that this algorithm can actively be used by anesthesiologists and surgeons for prevention of the complications connected with hypotonia such as postoperative myocardial infarction or sharp renal failure.

As Maxime Cannesson, the doctor of medical sciences, the leading researcher, professor of anesthesiology and the former head of the department of perioperatsionny medicine in the Medical center UCLA in Los Angeles noted in the statement, earlier doctors had no opportunity to predict hypotonia during transaction, and certainly, in such conditions anesthesiologists had to act very quickly in response to sudden falling of arterial blood pressure. The possibility of forecasting of episodes of hypotonia during transaction will allow doctors to prevent actively development of these episodes and their complications.[5]

AI better than doctors will recognize skin cancer

At the end of May, 2018 the research which showed more high efficiency of artificial intelligence in comparison with the person regarding recognition of cancer was published. However in hard-to-reach spots the computer is not so exact. Read more here.

Three most prospective applications of AI in medicine

Experts of Accenture analyzed the short-term value of medical solutions on the basis of artificial intelligence and selected three directions which have the largest potential in terms of financial profitability in the USA, Venturebeat reported on April 23, 2018.

Carrying out surgeries using robots is recognized the most cost-efficient. During similar transactions a series of small cuts, as a rule, becomes and miniature tools are used.

In this area several solutions are recognized perspective. So, cognitive surgical robotics allows to reduce duration of hospital treatment thanks to exact use of tools in each separate transaction depending on data of the patient. The complex for execution of operations of Da Vinci allows the surgeon to perform more effectively a number of difficult procedures, managing the robotic tool from the computer console. The miniature HeartLander robot allows to do heart operations through small cuts.

Experts considered the second perspective solution use of virtual assistants instead of nurses that allows to keep in contact of patients with medics and at the same time to reduce the number of appeals to hospitals. In Accenture gave the Sensely project developing virtual service of health service through mobile application which in 2016 attracted $8 million for development of the project as an example.

Automation of administrative document flow using AI became the third technology. First of all, it the solutions allowing to range urgent tasks and to save time on routine tasks, such as invoicing of recipes and analyses.

So, the products Nuance operate with clinical stories of patients and allow suppliers of medical services to save time on a reporting preparation. In the Clevelend clinic, the large private medical center of the State of Ohio, together with IBM the technology of support of medical solutions using the fast analysis of thousands of medical documents is implemented. GE Healthcare Camden Group implemented technology of processing of operational tasks (such as resource management of divisions and accommodation of patients) in the Maryland clinic of John Hopkins which is considered as one of the largest and modern medical centers of the world.

All listed solutions allow to reduce the probability of human errors and to increase efficiency of treatment. Service of program and technology complexes, their protection against failures and cyber attacks and also ensuring confidentiality of these patients becomes the main problem.[6]

The artificial intelligence was involved in ultrasonography diagnostics of pregnant women

The British hospital started a new type of testing of a fruit for pathologies which the doctor is not capable to notice. In the system based on artificial intelligence 350,000 pictures classified by these or those deviations[7] are put[8].

On Engineer refining, ultrasonography diagnostics with artificial intelligence received the name ScanNav and is intended to supply to the doctor with the additional information in real time. As a result of AI allows the specialist not to doubt that all foreshortenings are considered. The last is especially relevant because of the movement of a fruit in mother's womb.

So far the technology is approved in the test mode in obstetrics, but in the future development is going to be applied in the different fields of medicine. By the way, are laid already great hopes on AI diagnosticians in Japan having deficiency of doctors, and in China to artificial intelligence granted the medical license at all.

The artificial intelligence will be engaged in search of new antibiotics

Resistance to antibiotics is one of big problems of modern medicine. Thanks to universal application of antibiotics and nesoblyudeniye of instructions of the doctor of medicine ceased to influence bacteria that causes problems at treatment of both the most ordinary daily diseases, and heavy[9].

One equipment which can cope with resistance to antibiotics is a search of options of the known antibiotics. Unfortunately, it is extremely heavy and labor-intensive process requiring time. At least, for people. When go algorithms into action, the matter of time stops being so significant.

Group of the American and Russian researchers created an antibiotic algorithm which, quickly sorting databases, can open in 10 times more of options of antibiotics, than was open for all the time of similar researches in previous years.

The algorithm known as VarQuest, is described in article published in the last issue Nature Microbiology. Hossein Makhimani, professor of the university Carnegie-Mellon, says[10] in the press release that VarQuest completed search which would take with methods of traditional calculations hundreds of years.

Also Mokhimani specifies that VarQuest managed to provide more than one thousand options of the peptide groups used for production of antibiotics for record-breaking short time, and thus it can give to microbiologists wider perspective, perhaps, even to warn about trends or patterns of the microbiological world which differently would pass completely unnoticed.


In health care "tsunami" of AI technologies approaches

In the future the artificial intelligence (AI) will play a huge role in health care, is convinced Naveen Jain, the founder of the American startup of Viome specializing in medical technologies. The interview with the philanthropist and the innovator took place on fields of the international technology forum Slush 2017 which took place in Finland from November 30 to December 1, 2017. Read more here.

This AI tsunami approaches. Sensors become very cheap, and we can glance in an organism and precisely learn what in it occurs — he told CNBC, having added that the artificial intelligence will allow to analyze such amounts of data which people cannot process.

The AI device for remote control of a dream by means of radio waves is created

On August 8 it became known that engineers of the Massachusetts Institute of Technology (MIT) with the assistance of specialists of the Central hospital of the State of Massachusetts developed the AI system capable to control a dream of the person by means of radio waves.[11]

According to the TNW edition, the device which by the form reminds normal router Wi-Fi remotely analyzes radio signals around the person and determines dream stages by the movement of eyes — easy, deep or fast. As radio waves are reflected from a body, any small movement of a body changes the frequency of reflected modes. The analysis of these waves helps to reveal the vital parameters of life activity of the person, such as pulse and frequency of breath and to define aberrations. For functioning the device does not require sensors and is adapted for application in house conditions.

Imagine that your Wi-Fi-router knows when something dreams you, and can control whether it is enough to you time for a stage of a deep sleep that is necessary for recovery of normal work of memory" — Dina Katabi, professor of MIT heading researches noted.

It is supposed that monitoring of a dream in real time under natural conditions will allow to answer many questions connected with its frustration. As envisioned by scientific MIT, their development will turn as a result into the full-fledged tool which will allow attending physicians to trace dream parameters at distance, adjusting it in case of need.

The experiment on cloning of pigs in China was made by robots with AI

For the first time in the history of mankind the Chinese scientists from Institute of robotics and automated information systems at Nankai University of the city of Tianjin carried out successful cloning of pigs using robots, China People's Daily tells. At the beginning of January, 2017 510 cloned embryos were placed in six substitute sows. As a result of an experiment two sows at the end of April, for the 110th day of pregnancy gave birth to 13 healthy artificially brought pigs.[12]

When carrying out an experiment on cloning of pigs scientists for the first time used special robotic micromanipulators analyzers which executed all operations on collecting and transfer of DNA from donor animals to substitute carriers. Universal micromanipulators under control of artificial intelligence for transactions with DNA integrate in themselves functions of sampling of tests, testing and operating.

In the course of the cloning of pigs which is carried out in cooperation with Institute of livestock production and a veterinary research (Animal Husbandry and Veterinary Research Institute) the so-called technology of nuclear transfer of somatic cages (Somatic Cell Nuclear Transfer, SCNT) which is usually used for selection was involved — when the core of a somatic cage is transferred to an ovum without core. Advantage of this technique is the guarantee of high-quality insemination of an ovum, and a shortcoming — the low level of a successful completion of experiments because of big percent of defects in the course of cloning.

Authors of a research: professor Zhao Xin and his command. Photo:
Authors of a research: professor Zhao Xin and his command. Photo:

The main problem of process of cloning with nuclear transfer is in avoiding destruction of sensitive cages. Researchers made the preliminary analysis of the power necessary for the tool for safe work with cages during removal of cores, and then adjusted it at minimum possible level. Thanks to it extent of deformation of cages decreased from 30-40 mm to 10-15 mm that improved the subsequent development of a cage and increased chances of success.

It is supposed, the data on interrelation obtained as a result of a research microtransaction over cages and further development of cages will be able to help other scientists to make the following discoveries in this area.

Artificial intelligence taught to predict a heart attack better than doctors

In April, 2017 scientists from the University of Nottingham provided the artificial intelligence technology capable to predict approach of heart attack. Developers claim that forecasting accuracy is higher, than at doctors.

During the research compared efficiency of recommendations of physicians to work of four programs written using algorithms of machine learning. Scientists pursued the aim to find patterns in records more than 378 thousand patients. In the computer 22 criteria, including age, nationality, presence of arthritis and diseases of kidneys, cholesterol level in blood were put.

Scientists from the University of Nottingham provided the artificial intelligence technology capable to predict approach of heart attack.
Scientists from the University of Nottingham provided the artificial intelligence technology capable to predict approach of heart attack.

The conclusions drawn by artificial intelligence about risks of development of a heart attack verified with data for 2015, and they were more exact, than the predictions of doctors based on the recommendations of the American college of cardiology (American College of Cardiology, ACC) and the American association of heart (American Heart Association, AHA): from 74.5% to 76.4% of accuracy against 72.8%.

By approximate calculations of authors of the project, the computer could save 355 lives more, than a technique of ACC and AHA. Scientists intend to increase efficiency of an intelligent system, having added to it accounting of such risk factors as a way of life and genetic data.

It is interesting that algorithms did not consider influence of diabetes which was always considered as risk factor in the ACC and AHA system.

According to the epidemiologist of the University of Nottingham Stephen Weng, the biological systems have a set of interrelations which part is unknown to doctors: for example, the increased fat content in an organism under certain conditions can protect from sharp deviations in work of heart. Similar interactions are unevident, it is difficult to notice and explain them, but the computer program is capable to trace communication, having analyzed huge amounts of data, he considers.[13]

1970'e: Creation of the MYCIN system for diagnosing of bacteria

The Artificial Intelligence (AI) in the medical sphere began to be applied still in the seventies the last century when scientists of Stanford University on the basis of expert system of DENDRAL which is used in the field of organic chemistry created the MYCIN system for diagnosing of the bacteria which are catalysts of development of heavy infections — meningitis and bacteremia. Besides, the MYCIN system allowed to prepare individual recommendations about a dosage of antibiotics on the basis of the body weight of the patient. MYCIN is considered to be an example of the first use of artificial intelligence in medicine.

See Also