Artificial Intelligence in Medicine

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

Artificial intelligence in Russian medicine

• Artificial intelligence in Russian medicine

Directions of AI use in medicine

Development of new methods for disease prevention

AI can analyze medical data, including medical history, genetic data and lifestyle, to assess the risk of developing certain diseases, predicting the risks of various diseases, taking into account the state of health, genetic abnormalities, pathologies, living conditions, nutrition, etc.

It is possible not only to predict diseases, but also to identify groups of patients with a high risk of diseases, to organize preventive measures.

Medical Imaging and Diagnostics

Generative AI can be used to develop new diagnostic methods that can be more accurate and effective than existing methods, using accumulated disease and drug databases, combining, integrating and analyzing the best medical techniques and practices multiple times faster than a team of the most experienced physicians.

Artificial intelligence in radiology

Main article: Artificial intelligence in radiology

Generative AI models are able to improve the quality of medical images and provide a more accurate interpretation of the data. Such systems can automatically detect pathologies on X-rays, MRI or CT scans, which improves diagnostic accuracy.

Development of new drugs and treatments

Developing new drugs and treatments that may be more effective and safe than existing methods. For example, AI can be used to develop personalized drugs that can be tailored to each patient's individual characteristics, reducing the risks of side effects. AI can speed up the process of discovering and developing new drugs, using its algorithms to model and predict how molecules interact.

AI in Pharmaceuticals

Main article: AI in pharmaceuticals

Predicting epidemics

Using data on current infections, migration flows and climate change, AI can predict the spread of infectious diseases.

Tasks	Effect
Анализ (в т.ч. перекрестный) популяционных данных, данных ЕГИСЗ, омиксных данных, социальных сетей	Новые корреляции для дальнейшего научного исследования и применения в medicine
Анализ медицинских изображений, создание системы с автоматическим начальным уровнем описания и интерпретации результатов	Improving the speed and quality of medical decision-making
Умные скрипты опроса пациентов
СППВР* (по задачам, нозологиям), платформы организации СППВР как сервисов
Оперативный контроль качества и интеллектуальный бенчмаркинг оказания медицинской помощи в учреждении	Increase of speed and quality of control and expert work
Контроль отдаленных последствий оказания медицинской помощи	Change in the system of evaluation and analysis of health care delivery
Системы повышения приверженности граждан ЗОЖ и пациентов назначенному лечению	Reduction of morbidity and improvement of treatment effectiveness
Моделирование деятельности медицинской организации	Improved management, cost optimization
Носимые и иные мобильные медицинские изделия для дистанционного мониторинга	On-line/regular monitoring of health indicators
Умные учебные медицинские тренажеры	Improving the quality of training of healthcare professionals
Визуализация медицинских данных, вкл. умную навигацию при оперативных вмешательствах	Increase the speed and quality of medical decision-making, medical care

Standards in Artificial Intelligence in Healthcare

Main article: Standards in Artificial Intelligence in Healthcare

2024

How generative AI saves hundreds of billions of dollars to hospitals and clinics

By 2025, generative artificial intelligence (GenAI) will free up up to 10% of doctors' time, as well as help healthcare facilities save hundreds of billions of dollars. This is stated in the IDC study, the results of which are presented on January 23, 2024.

GenAI is becoming a transformative force in healthcare, analysts point out. One of the most promising options for using artificial intelligence and generative modeling is to accelerate the development of promising drugs. GenAI technologies can be used to create new protein sequences with specific properties designed for the development of antibodies, enzymes, vaccines, as well as gene therapy. In addition, GenAI tools allow you to analyze patient data and improve the quality of customer service by implementing a more personalized approach. AI helps engineers tailor instrument designs to specific patient needs and complex medical requirements, as well as accelerate the development process.

The IDC estimates GenAI technologies will deliver approximately $100 billion in annual health savings in the Asia-Pacific region alone by 2025. Globally, this figure will be much higher.

Several factors contribute to the growth of investment in GenAI in the medical sector, according to IDC. This is an increasing need for hyper-personalized patient care and the need to improve the quality of interaction. The use of AI-based diagnostics will lead to increased customer and staff satisfaction. Many health facilities will start using GenAI to address data fragmentation and improve workflows - this will help raise the level of security and efficiency of health care delivery.^[1]

8 AI applications that will revolutionize healthcare

In 2024, the impact of artificial intelligence (AI) technologies on healthcare will be deeper and larger than ever before. In January 2024, eight directions for the use of artificial intelligence in healthcare were presented, which will revolutionize the industry in a revolutionary way. He spoke about these directions on January 8, 2024, the information and analytical, educational project EverCare with reference to the portal Medium.com.

Direction No. 1. Predictive health monitoring

Wearable devices use AI algorithms to continuously track and analyze health metrics. They predict potential health problems before they become serious, alerting users and health professionals to do so. This proactive approach to disease prevention saves lives and cuts health care costs.

Direction No. 2. Personalized treatment plans

The era of universal treatments is a thing of the past. The ability of artificial intelligence to analyze vast amounts of data, including genetic information, lifestyle factors and previous health records, allows personalized treatment plans to be developed. These plans, made taking into account the individual characteristics of the patient, significantly increase the effectiveness of treatment, reduce side effects and improve the results of treatment.

Direction No. 3. Advanced Diagnostic Tools

AI-based diagnostic tools are more accurate and faster than ever. In 2024, AI is expected to help interpret complex medical images such as MRI and CT scans with an accuracy as good as experienced radiologists. These tools detect abnormalities earlier and with greater accuracy, leading to early intervention and improved prognosis for patients.

Direction No. 4. Robotic surgical care

The accuracy of using artificial intelligence in surgery has stepped far ahead. AI-enabled robotic systems provide surgeons with increased agility and control, making complex surgeries minimally invasive. These robotic assistants improve the results of operations, reduce recovery time and minimize the risk of infection.

Direction No. 5. Virtual Medical Assistants

medical assistants Virtual, powered by artificial intelligence, is a new medical support capability. They provide round-the-clock care by answering health-related questions, reminding patients to take medication and offering recommendations for minor health problems. Such a system of constant support is not only convenient for patients, but also eases the burden on medical institutions.

Direction No. 6. Drug Discovery and Development

As you know, AI accelerates the process of drug discovery and development - a process that traditionally takes years and costs a lot of money. By analyzing complex biochemical interactions, AI algorithms identify potential drug candidates in a fraction of the time. Such acceleration is critical for responding to emerging health crises and developing treatments for rare diseases.

Direction No. 7. Mental health monitoring and support

Mental health, often overlooked in traditional health care models, gets a much-needed boost from AI. Apps and platforms using AI algorithms enable early detection of mental health problems by analyzing speech patterns, text messages and even social media activity. These tools offer timely intervention and connect people with professional care, helping to address mental health issues before they escalate.

Direction No. 8. Improved training for health care professionals

At the same time, AI is revolutionizing the training of medical professionals. Virtual reality (VR) simulators created with AI provide a realistic and exciting learning environment. These simulations allow you to work out complex procedures without risk, giving medical professionals experience and confidence, while not putting patients at risk.

In conclusion, thanks to the integration of AI, the landscape of the healthcare industry in developed countries will change significantly in 2024. Not only does AI make healthcare more efficient and effective, it also makes it more accessible and personalized - from predictive health monitoring to advanced health professionals. As we continue to embrace these technological advances, AI's potential for further revolution in health care is limitless. The future has already arrived, and it is governed by AI.^[2]

2023

In the United States, shopping centers began to install the world's first AI offices of a doctor without doctors

In mid-November 2023, a startup in the region health care Forward introduced the CarePod platform - modular medical rooms based on artificial intelligence, in which appointments are carried out without the presence of a doctor. Such capsules are planned to be installed in shopping centers, gyms and office buildings. More. here

Neural network GPT4 began to be used in American health care

On August 18, 2023, the Langon Medical Center at the University of New York (NYU Langone) announced the start of using the large language model of GPT4 created by OpenAI. The neural network is used to increase the efficiency of various processes in the field of health care and reduce the burden on employees. Read more here.

Artificial intelligence taught to decipher thoughts via MRI

On May 1, 2023, American researchers at the University of Texas at Austin (UT Austin) announced the development of a new artificial intelligence-based system that can transform human brain activity into a continuous stream of text. A so-called semantic decoder in perspective can help people who have lost the ability to speak, for example, after a stroke, regain contact with the world.

The solution reportedly relies in part on a transformer model similar to that used ChatGPT in the company's chatbot. Unlike OpenAI other thought decoding systems under development, the new technology does not require surgical implants, making the process non-invasive. In addition, the proposed approach does not impose a limit on the number of words used.

Brain activity is measured using functional magnetic resonance imaging (fMRI). To teach the AI model, a person listens to podcasts or any audio recordings for a rather long time. After such training, the semantic decoder is able to generate text based on the patient's brain activity, say, when perceiving speech or thinking about an event. In other words, "reading thoughts" is made.

For the non-invasive method, this is a real leap forward from what was done before, when individual words or short sentences were usually used. We get a model for continuous decoding over long periods of time with complex ideas, "said Alex Huth, assistant professor of neuroscience and computer science at UT Austin.

It is noted that the result of the system is not a literal decryption. Instead, the semantic decoder allows you to capture the essence of what a person thinks about, who, for various reasons, is not able to express his thoughts in words.^[3]

2022

Global AI Systems Market for Medical Diagnostics Valued at $1.22 Billion

In mid-November 2022, Research And Markets announced the results of a study of the global market for artificial intelligence (AI) systems for medical diagnostics. Analysts predict that this industry will show steady growth.

The report says that the costs in the indicated area in 2022 will reach $1.22 billion. In the future, the CAGR (compound percentage CAGR) is expected to be 24.16%. Thus, by 2027, the market volume may increase approximately threefold - up to $3.60 billion.

The drivers of the industry are the introduction of AI tools in radiology and pathology; growing demand for intelligent systems against the background of optimized management and reduced human errors; improved and more accurate diagnosis of complex diseases. The market will also be boosted by cloud technologies and advances in intelligent image recognition. At the same time, several restraining factors stand out: higher cost compared to traditional means; the complexity of creating models and mechanisms for AI; issues related to cybersecurity and privacy.

Numerous sources, medical data such as,, MRI ULTRASONOGRAPHY mammography, genomics, computed tomography etc., are necessary for accurate detection of diseases using artificial intelligence technologies. According to experts, AI primarily improved the operation of hospital systems and accelerated the preparation of patients to continue their recovery at home.

The list of leading manufacturers of equipment and software for medical diagnostics using artificial intelligence includes:

• AMD;
• Aidoc Medical;
• AliveCor;
• Amazon Web Services (AWS);
• Babylon Healthcare Services;
• Deep Genomics;
• Enlitic;
• FDNA;
• General Electric;
• Google;
• HeartFlow;
• Intel;
• IBM;
• Koninklijke Philips;
• Medtronic;
• Microsoft;
• NVIDIA;
• Riverain Technologies;
• Siemens Healthineers.^[4]

Artificial intelligence taught to predict birth outcomes

On September 2, 2022, researchers from the Mayo Clinic reported that they had developed an artificial intelligence (AI) system that can predict birth outcomes. The technology has already begun to be used in clinical practice. Read more here.

Identifying fake COVID-19 vaccination certificates will help AI

Specialists from the International Institute of Engineering and Technology (IET) have published a study on the identification of fake certificates COVID-19 vaccination of technology-based bots (artificial intelligence AI) and deep learning. Their goal was programming Python to create a bot in a language into which one could simply download a certificate to check if it was fake or not. This became known on May 12, 2022.

To train their model, specialists used a large number of images of certificates (both genuine and fake) issued in the UK, USA, China, India, Dubai and Japan. To begin with, using local binary templates, the researchers removed all "noise" from the certificates, after which they extracted key elements from it, such as the logo, symbols, words and the crest-rough parameter for signature and printing. Next, using a convolutional neural network DenseNet201 they were able to determine the authenticity of the certificates.

In the functional block of the bot, experts included a deep learning model developed using Python Google in Collaborative, and hardware processor Intel Core Nvidia the following: i9-10,980HK, RTX 3080 GPU, RAM 32 GB storage and 21 TB.

According to the researchers, the model they presented with an accuracy of 0.94 exceeds the models for May 2022, including SVM, RNN, VGG16, Alexnet and CNN, in such performance indicators as accuracy, specificity, sensitivity, detection speed, completeness, F1 measure and calculation time^[5].

AI system sequenced human DNA in 5 hours

In mid-February 2022, a research team led by Stanford University set a new Guinness World Record for the fastest sequencing DNA human technique using computation AI to speed up workflow. More. here

2021

Investments in AI projects in the field of medicine in the world increased by $3 billion

In 2021, investments in artificial intelligence projects in the field health care around the world reached $11.2 billion against $8 billion a year earlier. Such data were released in March 2022 Stanford Institute artificial intelligence by the Stanford Institute for Human-Centered Artificial Intelligence (AI).

According to the study, in 2017-2021. medicine and healthcare became the most "attractive" industries for private investment in the artificial intelligence market. In total, more than $28.9 billion was invested in specialized projects during this period.

According to experts, thanks to artificial intelligence, the cost of bionic prostheses has decreased by 46.2% since 2017. If in 2017 such products cost an average of $42 thousand, then in 2021 the price of prostheses was already $22.6 thousand.

It is noted that simultaneously with the increase in investment in the industry, there was also an optimization of AI processes. So, from 2018 to 2021, the cost of software decoding of various medical images decreased by 63.6%, and the time of this process decreased by 94.4%. The researchers noted that the development of AI to analyze medical images positively affects the operational detection of neoplasms and various pathologies.

The report provides an example of the CVC-ClinicDB database, which is used to identify polyps in the kidneys. It consists of more than 600 high-resolution images obtained from 31 colonoscopies. Thanks to MSRF-Net's ultra-accurate neural network, image quality for CVC-ClinicDB has improved by 11.9% since 2015.

According to the researchers, the introduction of artificial intelligence technologies in medicine is one of the main trends in the world of health care. AI and neural networks are able to fundamentally change all world medicine: transform the diagnostic system, contribute to the development of new drugs, improve the quality of medical services in general and reduce costs.

AI Index Report 2022

AI Systems Market for Remote Patient Monitoring Valued at $900 Million

The global market for artificial intelligence technologies used in the process of remote monitoring of patients reached $893 million in 2021 against $712.7 million. This is evidenced by data from analysts at ResearchAndMarkets, released on December 22, 2021. Read more here.

Artificial intelligence improves IVF

By early December 2021, a group of startups had formed in the medical technology market, whose artificial intelligence technologies help improve IVF procedures. This topic is devoted to an article on the Bustle portal. Read more here.

Doctors start using AI diagnosing glaucoma in early stages with 97% accuracy

In early September 2021, doctors began using artificial intelligence that diagnoses glaucoma in the early stages, the diagnostic accuracy is 97%. Read more here.

WHO pointed out the negative consequences of the use of AI in medicine

At the end of June 2021, WHO pointed out the negative consequences of the use of artificial intelligence in medicine if its development, deployment and use are not based on ethical principles and the protection of human rights.

Like all new technologies, artificial intelligence can be misused and harm patients, "said WHO Director General Tedros Adhanom Ghebreyesus. To regulate and control the use of AI in medicine, WHO has published new recommendations that set out six principles for limiting risks and maximizing the use of AI's health capabilities.

The WHO report, "Ethical Principles and Application of AI for Health," states that AI can be used to accelerate and improve the accuracy of disease diagnosis and screening, assist in complex clinical situations, accelerate health research and drug development, and maintain a variety of public health interventions, including responses to infectious disease outbreaks and management of health systems. Artificial intelligence also allows patients to self-monitor their own health status and gives countries with limited resources the ability to access health services remotely.

However, experts warn doctors and patients against overly enthusiastic assessments of the health benefits of AI. WHO points to the possibility of unethical collection and use of health data, the manifestation of various biases of human communities in AI algorithms, and also recalls risks in relation to patient safety, cybersecurity and the environment. In addition, WHO notes that systems trained on data from high-income countries may not be effective in low- and middle-income countries.^[6]

Smart toilet designed to analyse chair

At the end of May 2021, an AI tool for analyzing the patient's stool was presented at Duke University. This technology will allow gastroenterologists to gather the information needed to properly treat chronic GI diseases such as inflammatory diseases and irritable bowel syndrome. Read more here.

2020

Medical AI solutions market reaches $4.2 billion - ResearchAndMarkets

The global market for artificial intelligence technologies used in healthcare reached $4.2 billion in 2020. This is evidenced by data from analysts at ResearchAndMarkets.

According to them, sales of medical AI solutions are growing intensively. So, until 2025 they will increase by 45.3% annually and will amount to $27.2 billion by the end of this period. At the same time, market dynamics in comparison with 2020 and 2019. experts do not lead.

Among the main growth drivers of the market in question, the researchers attributed:

• the growing body of medical data;
• increasing the complexity of datacets;
• the needs of healthcare organizations to reduce costs, including equipment, and increase computing power;
• an increase in the number of cross-sectoral partnership projects;
• an increasing imbalance between the number of doctors and patients, which is pushing the development of demand for improvised medical services.

In addition, the rise of the medical AI solutions market is facilitated by the introduction of such technologies by numerous pharmaceutical and biotechnology companies around the world. They, in particular, use such developments to create vaccines and drugs for the treatment of coronavirus COVID-19.

As for the negative factors restraining the development of the market for AI solutions for healthcare, analysts consider such to be the reluctance of practicing doctors to introduce artificial intelligence into work, a shortage of specialists capable of working with such technologies and ambiguous legislative regulation of the medical software market.

Also, the market is hampered by the lack of carefully selected medical data, concerns about the confidentiality of such information and the problem of integration between AI solutions from various manufacturers.

Artificial intelligence in healthcare is by far one of the most important scientific advances in medicine. The involvement of several startups in the development of imaging and diagnostic solutions based on artificial intelligence is a key factor contributing to the growth of the sector, the study said, excerpts from which were published in April 2021.

Artificial intelligence has great potential in senior care projects, genome research, new drug discovery, medical imaging and diagnostics of various diseases, analysts said.

In 2020, there were several, according to analysts, notable events in the market for AI solutions for healthcare. For example, IBM Watson Health and EBSCO Information Services entered into a contract to provide access to evidence-based information about drugs and diseases that can help doctors and patients cope with infectious diseases, including COVID-19.

Another eye-catching deal is between GE Healthcare and South Korea's startup Lunit. The companies launched a complex for chest X-ray analysis based on artificial intelligence.

The following companies were named the largest developers of medical AI tools:

• Nvidia;
• Intel;
• IBM;
• Google;
• Microsoft;
• General Electric X-ray (GE Healthcare);
• Siemens Healthineers;
• Medtronic;
• Micron Technologies;
• Amazon Web Services;
• Johnson & Johnson;
• Philips;
• General Vision Services;
• Cloudmex;
• Oncora Medical;
• Anju Life Sciences Software;
• CareSkore;
• Linguamatics;
• Enlitic;
• Lunit;
• CureMetrix;
• Qure.ai Technologies Private;
• Context Vision Operations;
• Caption Health;
• Butterfly Network;
• Imagia Cybernetics;
• Precision Health Intelligence;
• Cota Healthcare;
• FDNA;
• Recursion Pharmaceuticals;
• Atomwise;
• Deep Genomics;
• Cloud Pharmaceuticals;
• Welltok;
• Vitagene;
• Lucina Health;
• Next IT;
• Babylon Health;
• MDLIVE;
• Magnea;
• Physiq;
• CyrcadiaHealth;
• Caresyntax.^[7]

Announcement of an AI system for predicting risks in pregnant women

In early September 2020, researchers at Carnegie Mellon University presented a machine learning technique that allows you to analyze placental samples and calculate a woman's health risk in future pregnancies. The system is designed to help obstetricians-gynecologists, who will be very useful for the forecast of possible complications of future women in labor. According to the authors of the project, their development has already begun to be used in clinical practice. Read more here.

Microsoft invests $40 million in AI for healthcare

On January 30, 2020, Microsoft announced the launch of the five-year AI for Health program, in which it will invest $40 million in artificial intelligence (AI) technologies for healthcare over five years. Read more here.

2019

Artificial intelligence began to be used to predict the need for hospitalization

In November 2019, it became known that in medical institutions began to use artificial intelligence to predict the need for hospitalization of patients. This technique has become available in clinical practice with the support of the British Ministry of Health.

The project uses a system algorithm to analyse the complexity of the health status of sufferers, predict which patients may need to be admitted to hospital and help GPs work to reduce this risk. So, doctors could know in advance when to intervene to help doctors make accurate forecasts, it is better to plan the use of resources - the number of hospital beds, medicines, etc.

It is an analysis in the form of difficulty measures based on a percentage scale. The assessment is related to major health indicators and other factors such as high blood pressure, sedentary lifestyle or past smoking. For example, a patient with a complexity score of 80% would have a high risk of needing hospitalization.

According to the British Ministry of Health, artificial intelligence in the field of health care is an opportunity to check your own health, just as a credit history is checked. And the new development, which has begun to be applied in several medical institutions, shows how the application of modern technologies can have a positive impact on patient care.

The department is convinced that the new project is able to help transform the health care system. After all, a better understanding of the needs of patients and more effective planning also have an impact on funding - it is much cheaper to provide medical care on the spot than to take a patient to a hospital.^[8]

System for predicting epileptic seizure

In mid-November 2019, an artificial intelligence system was presented that can predict epileptic seizures with an accuracy of 99.6%. Moreover, it is able to predict their development an hour before the onset of the main symptoms. This precise prognosis allows people to prepare for an attack in time and take medication. The solution has already begun to be used in practice. Read more here.

The medical center began to use a facial recognition system to detect genetic abnormalities in newborns

In mid-November 2019, it became known that Chinese scientists developed and implemented an artificial intelligence-based facial recognition system designed to detect genetic abnormalities in newborn screening. Read more here.

Doctors start predicting ECG death

In mid-November 2019, AI technology was introduced that can predict heart rhythm failures and accurately predict the risk of death in patients, even when independent cardiologists cannot recognize the same risk factors. Read more here.

CB Insights: Medical AI Technology Market to Reach $6.6 Billion in 2021

At the beginning of 2019, according to the analytical company CB Insights, starting in 2013, international technology startups developing artificial intelligence technologies managed to raise $4.3 billion in 576 transactions. In addition, experts say that over the next three years, the market for medical AI technologies will reach $6.6 billion, increasing by 40% each year.

IBM and AstraZeneca have created a neural network that foreshadows a heart attack

In early March 2019, IBM and AstraZeneca unveiled a neural network that can predict a heart attack. The results of the new technology are described in the published article "Clustering based on the results of patients with acute coronary syndrome when using a multitasking neural network."

The team of researchers collected data on age, sex, history of life and disease, bad habits, as well as laboratory results, information on the treatment being conducted and almost 40 other indicators among 26,986 adult hospitalized patients in 38 urban and rural hospitals in China. All data were uploaded to the neural network, which was supposed to find out whether the patient had a serious adverse cardiac event (MACE) in the past, as well as whether he received antiplatelet drugs, beta-blockers and statins - drugs that reduce the manifestations of coronary insufficiency and prevent myocardial infarction and stroke.

Further, the authors of the article performed k-mean clustering to distribute patients into seven groups based on data obtained by the neural network. As a result, it turned out that in the first cluster, which contained patients with frequent cardiovascular events by type of heart attack and stroke, but low incidence of coronary heart disease, the main predictor of the next heart attack was the presence of diabetes mellitus, while in another cluster that included patients with a severe course of cardiovascular pathology without a prior infarction, the main predictors were older age and elevated systolic blood pressure.

The researchers caution that while clustering has implications for disease prognosis, it is unclear whether these data can be used effectively in clinical practice. However, their work demonstrates that AI-based cluster analysis is a promising approach for classifying patients with myocardial infarction. Future studies will focus on identifying "cluster-specific" interventions that consider efficacy.^[9] of prior treatment.

2018

The volume of the market for AI technologies in healthcare amounted to $1.4 billion - Zion Market Research

In 2018, the global market for AI technologies for healthcare reached $1.4 billion, according to the analytical company Zion Market Research. It is expected that by 2025 the figure will grow to $17.8 billion, and the cost of such solutions will increase by about 43.8% annually.

Most of all they spend on medical artificial intelligence (machine learning, context-sensitive computing, natural language processing, computer vision, speech recognition) in North America. The leadership comes as the region is represented by tech giants such as Microsoft, IBM, Google, Nvidia, Amazon, Intel, General Electric and Xilinx. In addition, mergers and acquisitions, large partnerships and the launch of important products are frequent in North America.

In Europe, by 2019, the market for artificial intelligence used for medical purposes can be considered nascent. In 2016, its volume was measured at $320 million, by 2019 it will amount to $1.61 billion. At the same time, 21% of medical institutions in Europe plan to purchase AI tools, according to data from the European e-health community, released in April 2019.

One of the main catalysts for demand for AI products in medicine is the shortage of doctors. According to the World Health Organization, by 2019, 57 countries lack approximately 2.3 million nurses and doctors. A factor restraining the development of this market, experts call the lack of qualified specialists who could follow the guidelines in the field of AI.^[10]

Among the largest manufacturers of AI solutions, analysts include the following companies:

• Google;
• GE Healthcare;
• Intel;
• Medtronic;
• Siemens Healthineers;
• General Vision;
• Amazon Web Services (AWS);
• Nvidia;
• IBM;
• AiCure;
• iCarbon;
• Cyrcadia Health;
• Atomwise;
• Pathway Genomics;
• Zebra Medical Vision AI1 (All-In-One);
• Sophia Genetics;
• Apixio;
• Microsoft.

Artificial intelligence is presented, increasing the success of IVF by 20%

At the end of December 2018, experts from the University of Cornwall in the USA and Imperial College in London demonstrated the results of their study, according to which the effectiveness of IVF can be increased by 10-20% if artificial intelligence is used to assess the quality of embryos. Read more here.

Start of installation in China of 4 thousand booths with AI doctors diagnosing in minutes

At the end of November 2018, the largest online health care provider in China, Ping An Healthcare and Technology, spoke about the launch of a project to install several thousand AI clinics the size of a phone booth. It is planned to distribute them throughout the country in three years. The first such points of medical care have already started working. Read more here.

How artificial intelligence will evolve in medicine in 2019

In November 2018, DataArt, which specializes in consulting and IT solution development services, presented a forecast of how artificial intelligence will develop in medicine in 2019.

According to experts, artificial intelligence will remain an object of interest for both investors and medical workers. AI algorithms are still evolving, becoming faster and more accurate. At the same time, only a few pharmaceutical companies have integrated solutions based on artificial intelligence technologies into their processes. In most cases, such solutions are used only in pilot projects and have not yet received proper deployment. Health care in 2019 is waiting for progressive and non-standard views that will show how to fully use all the possibilities of AI.

Thanks to artificial intelligence, "smart" telemedicine services will make quality medicine more accessible to a wide range of people and will help them prevent the development of chronic diseases through timely consultations with the doctor.

Analysts are confident that algorithms related to the collection, processing and storage of data in 2019 will be of great interest and importance for the healthcare industry. With the use of next-generation sensors, continuous monitoring of vital patient health indicators has already become a reality.

A modern medical diagnostic examination provides far more detail than it did 30 years ago. It includes data from different sources - from family history to protein concentration in a blood sample. 

Data from mobile devices creates a dense stream of data that needs to be processed and stored, and 2019 should be the year when progress in this direction will increase.^[11]

Japan builds AI hospitals to tackle doctor shortage

In August 2018, it became known that the Japanese government, with the support of business and the scientific community, is starting to build hospitals in the country in which artificial intelligence will come to the aid of doctors. At the expense of AI technologies, it is planned to cope with the shortage of doctors in Japan, relieve staff and reduce medical costs. Read more here.

The first recommendations on the use of AI in the field of health care are proposed

On June 18, 2018, the American Medical Association (AMA) proposed the world's first recommendations for the use of artificial intelligence in health care. The statement, which an AMA spokesman announced at the annual conference in Chicago, outlined the main directions for the further development of AI in this industry.

According to this statement, the AMA intends to implement developments in artificial intelligence and other priority areas to improve treatment outcomes and to professionally satisfy doctors. The AMA is going to use its meaningful position in the industry to engage manufacturers, prioritize AI development, and address validation and implementation challenges. In addition, the AMA intends to develop a plan to train specialists and communicate information to patients about the limitations and opportunities that are characteristic of this category of analytical tools.

The AMA advocates the integration of elaborate, high-quality, and clinically proven AI practices, and requires proper professional and government oversight of their safe, effective, and legal use. AI-based analytical technologies, the AMA believes, should be available to verify and identify systematic errors at all stages of development, meet leading standards of reproducibility, and protect the interests of individuals and the confidentiality of personal information.

The AMA believes that the focus should be on the needs of users, and the use of the AI system should be tested on a representative sample as part of a clinical trial.

The combination of AI methods and the irreplaceable experience of the clinician will undoubtedly improve the outcomes of therapy, - said AMA board member Jess M. Ehrenfeld. "However, we must be directly involved in solving all problems that arise in the design, assessment and implementation of these methods, because every year their application is becoming wider.[12]

AI taught to predict a drop in blood pressure during surgery

In June 2018, the journal Anesthesiology published results from a team of researchers who developed an algorithm to predict potential hypotension or an abnormal drop in blood pressure during surgery.

To create the algorithm, the researchers used machine learning technology - artificial intelligence analyzed data from 1,334 patients, during the operation of which blood pressure was recorded - a total of 545,959 minutes. Based on these data, an algorithm for predicting hypotension during surgery was prepared.

Having approved this algorithm, the researchers tested it on a second dataset, which included blood pressure indicators of 204 patients with a total duration of 33,236 minutes. These records included 1923 episodes of hypotension. The algorithm accurately predicted a sudden drop in blood pressure 15 minutes before it occurred in 84% cases, 10 minutes before it occurred - in 84% cases and five minutes before it appeared - in 87% cases.

The researchers suggest that this algorithm may be actively used by anesthesiologists and surgeons to prevent complications associated with hypotension, such as postoperative myocardial infarction or acute renal failure.

As Maxim Cannesson, MD, PhD, lead researcher, professor of anesthesiology and former chair of perioperative medicine at UCLA Los Angeles Medical Center, noted in a statement, doctors previously had no way to predict hypotension during surgery, and of course, in such conditions, anesthesiologists had to act very quickly in response to a sudden drop in blood pressure. The possibility of predicting episodes of hypotension during surgery will allow doctors to actively prevent the development of these episodes and their complications.^[13]

AI is better than doctors at recognizing skin cancer

At the end of May 2018, a study was published that showed higher efficiency of artificial intelligence compared to humans in terms of cancer recognition. However, in hard-to-reach places, the computer is not so accurate. Read more here.

The three most promising applications of AI in medicine

Accenture experts analyzed the short-term value of AI-based medical solutions and identified three areas that have the greatest potential in terms of financial profitability in the United States, Venturebeat reported on April 23, 2018.

Performing surgical operations using robots is recognized as the most cost-effective. During such operations, as a rule, a series of small cuts are made and miniature tools are used.

Several solutions are recognized as promising in this area. Thus, cognitive surgical robotics allows you to reduce the duration of inpatient treatment due to the exact use of tools in each individual operation, depending on the patient's data. The Da Vinci Operations Complex allows a surgeon to perform a number of complex procedures more efficiently by controlling a robotic tool from a computer console. The miniature robot HeartLander allows you to perform heart surgery through small incisions.

The second promising decision, experts considered the use of virtual assistants instead of nurses, which allows supporting communication patients with health workers and at the same time reducing the number of visits to hospitals. As an example, Accenture cited the Sensely project, which develops a virtual medical service through, mobile application which in 2016 raised $8 million for the development of the project.

The third technology was the automation of administrative document management using AI. First of all, these are solutions that allow you to rank urgent tasks and save time on routine tasks, such as writing prescriptions and analyzes.

Thus, Nuance products operate on clinical stories of patients and allow medical service providers to save time on reporting. Cleveland Clinic, a large private Ohio medical center, together with IBM has implemented technology to support medical solutions by quickly analyzing thousands of medical documents. GE Healthcare Camden Group has implemented operational task processing technology (such as unit resource management and patient placement) at Maryland's Johns Hopkins Clinic, considered one of the world's largest and most advanced medical centers.

All of these solutions can reduce the likelihood of human error and increase the effectiveness of treatment. The main problem is the maintenance of software and technology complexes, their protection from failures and cyber attacks, as well as ensuring the confidentiality of patient data.^[14]

2017

A "tsunami" of AI technologies is coming in healthcare

In the future artificial intelligence , (AI) will play a huge role in healthcare, said Naveen Navin Jane Jain, founder of Viome an American startup specializing in medical technology. An interview with a philanthropist and innovator took place on the sidelines of the international technology forum Slush 2017, held in Finland from November 30 to December 1, 2017. More. here

A real AI tsunami is coming. Sensors are becoming very cheap, and we can look inside the body and find out exactly what is happening in it, "he told CNBC, adding that artificial intelligence will allow analyzing such volumes of data that people cannot process.

AI device for remote sleep control using radio waves has been created

On August 8, it became known that engineers at the Massachusetts Institute of Technology (MIT), with the participation of specialists from the Massachusetts Central Hospital, developed an AI system capable of controlling human sleep using radio waves.^[15]

According to TNW, a device that looks like an ordinary one router Wi-Fi remotely analyzes radio signals around a person and determines the stages of sleep by eye movement - light, deep or fast. Since radio waves are reflected from the body, any small movement of the body changes the frequency of reflected waves. Analysis of these waves helps to identify vital parameters of human vital activity, such as pulse and respiratory rate, and determine abnormalities. For operation, the device does not require sensors and is adapted for use at home.

Imagine that your Wi-Fi router knows when you are dreaming of something and can control whether you have enough time for the deep sleep stage, which is necessary to restore normal memory function, "said Dina Katabi, MIT professor who led the research.

It is assumed that real-time sleep monitoring in natural conditions will answer many questions related to his disorder. As conceived by MIT scientists, their development will eventually turn into a full-fledged tool that will allow attending physicians to track sleep parameters at a distance, adjusting it if necessary.

China pig cloning experiment carried out by AI robots

For the first time in human history, Chinese scientists from the Institute of Robotics and Automated Information Systems at Nankai University in Tianjin have successfully cloned pigs using robots, China People's Daily reports. In early January 2017, 510 cloned embryos were placed in six surrogate sows. As a result of the experiment, two sows at the end of April, on the 110th day of pregnancy gave birth to 13 healthy artificially bred piglets.^[16]

When conducting an experiment on cloning pigs, scientists for the first time used special robotic micro-manipulators-analyzers, which performed all operations to collect and transfer DNA from donor animals to surrogate carriers. Universal micromanipulators controlled by artificial intelligence for DNA operations combine the functions of sampling analyzes, testing and operation.

In the process of cloning pigs, carried out in collaboration with the Institute of Animal Husbandry and Veterinary Research Institute, the so-called Somatic Cell Nuclear Transfer (SCNT) technique, commonly used for selection, was used when the somatic cell nucleus is transferred to a nucleus-free egg cell. The advantage of this technique is the guarantee of high-quality insemination of the egg, and the disadvantage is the low level of successful completion of experiments due to the large percentage of rejection during the cloning process.

The main challenge of the nuclear transfer cloning process is to avoid the destruction of sensitive cells. The researchers made a preliminary analysis of the power needed by the tool to safely work with cells when removing nuclei, and then adjusted it to the lowest possible level. Thanks to this, the degree of cell deformation decreased from 30-40 mm to 10-15 mm, which improved the subsequent development of the cell and increased the chances of success.

It is assumed that the data obtained as a result of the study on the relationship between microoperation over cells and further cell development will be able to help other scientists make the following discoveries in this area.

Artificial intelligence taught to predict heart attack better than doctors

In April 2017, scientists at the University of Nottingham unveiled artificial intelligence technology capable of predicting the onset of a heart attack. The developers argue that the prediction accuracy is higher than that of doctors.

The study compared the effectiveness of medical recommendations with the work of four programs written using machine learning algorithms. Scientists pursued the goal of finding patterns in the records of more than 378 thousand patients. The computer included 22 criteria, including age, nationality, the presence of arthritis and kidney disease, and blood cholesterol.

The conclusions made by artificial intelligence about the risks of developing a heart attack were compared with the data for 2015, and they turned out to be more accurate than the predictions of doctors based on the recommendations of the American College of Cardiology (ACC) and the American hearts Heart Association (AHA): from 74.5% to 76.4% accuracy versus 72.8%.

According to rough estimates by the authors of the project, the computer could have saved 355 more lives than the ACC and AHA methods. Scientists intend to increase the efficiency of the intellectual system by adding to it the consideration of such risk factors as lifestyle and genetic data.

Interestingly, the algorithms did not account for the impact of diabetes, which has always been considered a risk factor in the ACC and AHA system.

According to University of Nottingham epidemiologist Stephen Wang, biological systems have many relationships, some of which are unknown to doctors: for example, increased body fat under certain conditions can protect against acute abnormalities in the heart. Such interactions are not obvious, they are difficult to notice and explain, but the computer program is able to trace the connection by analyzing huge amounts of data, he said.^[17]

2016: Microsoft develops artificial intelligence to fight cancer

In September 2016, Microsoft announced the development of artificial intelligence, which, as the company expects, will help doctors find the right methods of treating cancer. This research project was named Hanover. Read more here.

1970's: Building a MYCIN system to diagnose bacteria

Artificial intelligence (AI) in the medical field began to be used back in the 70s of the last century, when Stanford University scientists, based on the DENDRAL expert system used in the field of organic chemistry, created the MYCIN system for diagnosing bacteria that are catalysts for the development of severe infections - meningitis and bacteremia. In addition, the MYCIN system made it possible to prepare individual recommendations for the dosage of antibiotics based on the patient's body weight. MYCIN is considered to be an example of the first use of artificial intelligence in medicine.

Notes