December 21 2022
Reading time 17 minutes
Artificial intelligence in health care is of great interest for many specialists: from health professionals and investors to scientists and doctors. In this article we will describe what is being tested and applied, and what challenges face digital evolution.
Why are technologies needed in health care?
Predicting the clinical outcome. Models examine clinical data and predict complications and death rate. For instance, in case of misleading mild start of COVID-191 or high risk of cancer2.
Saving time . AI helps a doctor to find necessary information using electronic medical recordsу3. Mobile apps for hospitals reduce time spent in queues. They ask questions and make a route to a doctor’s office4.
Saving resources in health care. Manual work is automated, which allows a doctor to increase admissions. AI monitors a patient’s condition, reduces the number of hospital admissions an improves quality of examinations5.
Digital solutions in health care can be both virtual and physical. Virtual ones include electronic medical records, neural networks and mobile apps. Physical solutions include robotic systems6.
Machine Learning (ML) is a branch of artificial intelligence (AI) that uses data and algorithms to learn how to solve problems and improve the accuracy of solutions. 7
The technology combines statistical methods and computer science, enabling data processing by computational algorithms. The goal is to find valuable dependencies in order to diagnose or choose a treatment method 8. There are two areas of ML:
Supervised Learning. A training sample is a set of data labeled by a human in order to be used by a computer. After obtaining new information, AI compares it with the existing knowledge and then classifies the result 9. For instance, the technology finds a suspicious neoplasm on an X-ray image and identifies it as a lung nodule8.
Unsupervised learning. The machine does not know what the predicted result should be. Instead, it searches for hidden patterns in the data. This is how diseases with numerous and complex causes are analyzed 8.
To communicate with a human, AI uses Natural Language Processing (NLP). This information technology recognizes oral or written messages, breaking them down into separate units of language — words or sentences. NLP learns how to understand the context, such as medical terms 10.
Modern technologies are used in various fields of medicine. For example, if a patient has a cold, they help a doctor to analyze an X-ray image of the patient’s lungs, find the results of the laboratory tests and identify the source of infection. AI applications in a hospital are summarized below.
An electronic medical record (EMR) comprises various pieces of information about a patient 11:
demographic data, such as gender and age
wordings of diagnoses
laboratory test results
CT and MRI scans
marks about surgeries.
This digital alternative to a paper record is becoming a source of knowledge for AI. EMR implementation is a strategic area aimed to 12:
create a system to automate doctors’ routine
offer personal treatment plans
improve quality and safety of medical care
develop medicines using information technologies.
EMR Pneumonia by SberMedAI analyzes clinical data of patients with pneumonia. The system predicts the likelihood of complications. Based on EMR Pneumonia reports, the attending physician chooses further treatment tactics.
When applying EMR, data are kept secure and confidential. Information is transmitted via secure communication channels 13.
However, the prospects of electronic document management in healthcare are not flawless. Scientists emphasize that it is critical to organically integrate EMR into a doctor-patient relationship, preserving communication as the basis of clinical thinking 14.
There was a study which estimated that emergency department personnel made 4,000 mouse clicks during a 10-hour shift, with 44% of the working time spent to enter clinical information into the computer. The authors believe that EMR development opportunities include improved data entry 15.
Artificial intelligence has given a second wind to using X-ray for diagnosing respiratory diseases. Information technologies can 16:
calculate the volume and density of lung tissue
assess the structure of the respiratory tract
identify signs of acute and chronic diseases of the respiratory system
determine the degree of damage in COVID-19 pneumonia
study changes in the lungs by comparing the new image with the old one.
СT Lungs reduces the burden on radiologists by accelerating the performance of routine tasks: it helps to detect lung tissue damage locations, calculate the volume and degree of damage and form a preliminary description of the examination results.
AI can also be used for retrospective diagnosis. Lung RSA solution analyzes the archive of CT scans collected during the COVID-19 pandemic. The process is as follows:
The algorithm searches for signs of malignant neoplasms, e.g. lung tissue nodules.
Information about suspicious findings is sent to the MDDC reference center or directly to the regional cancer center.
The AI-generated preliminary medical opinion is then reviewed and verified by a doctor.
The service has been implemented in the Nizhny Novgorod Regional Clinical Oncology Center, whose MIS has been directly connected to the MDDC platform, enabling prompt data transmission and processing. The specialists of the center received 125 lung CT scans for verification, on which AI revealed nodules.
Information technologies help reduce radiation dose during X-ray examinations. AI performs calculations and suppresses noise on the original X-ray image, generating an image of higher quality. The computer system targets the X-ray tube more accurately. Proper targeting helps to avoid uninformative X-rays 17.
AI analyzes CT scans of the brain in order to reveal signs of stroke. CT Stroke processes images to highlight areas of acute impairment of cerebral circulation. An additional algorithm can be used to rate changes based on the ASPECTS score.
CT Stroke can be applied in conjunction with CT Angiography. The services complement contrast-enhanced computed tomography, helping radiologists to detect blood clots.
Another application of the "second vision" in medicine is dermatology. Technologies analyze 18:
images obtained using a microscope.
Upon identification of a suspicious area, the computer assistant classifies it. Diagnoses can be different: from eczema or psoriasis to malignant neoplasms, such as melanoma 18.
Information technologies have been applied in diabetic retinopathy screening. Such healthcare solutions include RetmarkerDR (Portugal) and EyeArt (USA). The algorithm analyzes images of the ocular fundus, providing ophthalmologists with various pieces of information20:
suggesting whether there is a disorder
assessing its severity
detecting microaneurysms, or enlarged areas of capillaries
detecting retinal hemorrhages
comparing the image with the previous ones to assess the dynamics.
The use of AI improves breast cancer screening. The machine analyzes a digital mammogram to find suspicious areas and highlights them with a frame. Then the algorithm calculates the probability that the neoplasm is malignant 21.
The Mammography service by SberMedAI is an example. The algorithm reduces the burden on a doctor by automatically marking images and rating changes based on the BI-RADS system. It increases the speed and accuracy of screening, allowing early disorder identification.
Information technologies are also used in cardiology. Applications include cardiovascular disease screening based on ECG records. The neural network learns from electrocardiograms examined by doctors. They can be downloaded from different devices 22:
implantable loop recorders
an AI-driven stethoscope
a portable ECG monitor.
MDDC Cardio by SberMedAI is a complex combining the Valenta telecardiology unit and artificial intelligence. Automatic decryption using AI enables faster and more accurate analysis of ECG recordings. The result is transmitted to the EMR. A health professional can open a preliminary AI report and verify it.
Emergency medical stations and paramedic and obstetric stations in the Nizhny Novgorod Region have been equipped with telecardiology units, with 638 devices connected to a single network. The service has already helped doctors to identify 6,143 cases of acute coronary syndrome and myocardial infarction.
A brief description of how it works is given below 23:
An ECG signal is converted into a sequence of digits.
Input data is sent to the neural network.
At each intermediate level, it makes a decision.
As a result, a preliminary medical opinion is generated.
Information technologies recognize arrhythmias and changes in the work of heart chambers — atria or ventricles. AI assistance is critical in the diagnosis of life-threatening conditions, such as asymptomatic atrial fibrillation 24.
The software is installed on mobile devices. One can quickly conduct screening and then transfer the data via smartphone. This approach saves resources of the health care system, and developers are constantly improving accuracy of the results 24.
Each infection is caused by a particular pathogen, for example, a bacterium or virus. At the early stages of the disease, it is not always revealed, so it is impossible to prescribe specific treatment. It takes time and effort to identify the pathogen in a laboratory 25.
Information technologies helps quickly find the disguised cause of the disease and choose the appropriate medication. AI analyzes images obtained from a digital microscope. The main task is to classify the pathogen according to microbiological "evidence"26:
interaction of viral particles with body proteins
changes in the structure of affected cells
presence and number of pathogens.
Algorithms analyze the genome sequence of pathogens such as COVID-19 or Ebola, allowing quicker understanding of the origin and structure of the virus and its role in the disease by scientists. This accelerates the development of vaccines and antiviral drugs. 27
In some cases, biopsy may be taken to accurately establish a diagnosis. The obtained tissue fragment is examined under a microscope. AI searches for changes in cells, surrounding tissue, and blood vessels. Information technologies help to distinguish between benign and malignant diseases, as well as to predict the development of cancer 28.
The relevance of AI in surgery us related to the increased cognitive load on doctors who need to perform a lot of manual manipulations. The purpose of special systems is to monitor the progress of the operation and assist the surgeon when needed. Algorithms learn how to analyze 29:
preoperative images, e.g. CT scans
indications of medical devices, such as an anesthetic monitor
the results of studies during surgery — endoscopy or ultrasound.
This field requires expertise, so the training data is annotated by doctors 23. The system learns how to determine the location of the tumor, select candidates for surgery, and predict tissue damage 30.
The use of navigation systems helps the surgeon to distinguish between affected and normal tissues. The technology combines images of anatomical structures, such as CT scans. The algorithm displays a three-dimensional view of organs and blood vessels on the monitor 31.
Robotic assistants found their application in medicine. Da Vinci, a surgical system developed by the American company Intuitive Surgical, has recently appeared in hospitals. A surgeon works in tandem with AI to perform complex operations with minimal tissue damage.
The system is managed via the console: the surgeon remotely controls the instruments. The robot’s limbs move like a human hand, but provide a greater range of motion. The system creates high resolution 3D images of anatomical structures 32.
Just like doctors, computers need to learn. Developers carefully select "textbooks" so that AI makes right decisions. At that, they face the following data use difficulties 33:
heterogeneity: data comes from different sources and in different formats
limitation or unavailability: in the case of rare diseases, it takes many years to collect information in order to obtain accurate results
inadequacy: data does not always objectively represent the problem.
A lot of time is needed to implement an information system in health care practice. A developed model is being tested and proves its effectiveness in the chosen field. Then the new solution is gradually integrated into the work of a doctor 34.
The use of technologies in health care can lead to biased decisions as data is influenced by sociological and economic factors. Some models are organized according to the "black box" principle, making it difficult to understand how the conclusion was made 35.
The digital evolution in health care is inspired by technology, but the main role is played by doctors, who act as active users. Taking into account the conclusions made by the machine, they still have to make their own clinical decisions.
Ko H, Chung H, Kang WS, Park C, Kim DW, Kim SE, Chung CR, Ko RE, Lee H, Seo JH, Choi TY, Jaimes R, Kim KW, Lee J. An Artificial Intelligence Model to Predict the Mortality of COVID-19 Patients at Hospital Admission Time Using Routine Blood Samples: Development and Validation of an Ensemble Model. J Med Internet Res. 2020 Dec 23;22(12):e25442. doi: 10.2196/25442. PMID: 33301414; PMCID: PMC7759509.
Yeh MC, Wang YH, Yang HC, Bai KJ, Wang HH, Li YJ. Artificial Intelligence-Based Prediction of Lung Cancer Risk Using Nonimaging Electronic Medical Records: Deep Learning Approach. J Med Internet Res. 2021 Aug 3;23(8):e26256. doi: 10.2196/26256. Erratum in: J Med Internet Res. 2021 Oct 15;23(10):e33519. PMID: 34342588; PMCID: PMC8371476.
Chi EA, Chi G, Tsui CT, Jiang Y, Jarr K, Kulkarni CV, Zhang M, Long J, Ng AY, Rajpurkar P, Sinha SR. Development and Validation of an Artificial Intelligence System to Optimize Clinician Review of Patient Records. JAMA Netw Open. 2021 Jul 1;4(7):e2117391. doi: 10.1001/jamanetworkopen.2021.17391. PMID: 34297075; PMCID: PMC8303101.
Li X, Tian D, Li W, Dong B, Wang H, Yuan J, Li B, Shi L, Lin X, Zhao L, Liu S. Artificial intelligence-assisted reduction in patients’ waiting time for outpatient process: a retrospective cohort study. BMC Health Serv Res. 2021 Mar 17;21(1):237. doi: 10.1186/s12913-021-06248-z. PMID: 33731096; PMCID: PMC7966905.
Socio-Economic Impact of AI on Healthcare [Electronic resource]: Deloitte. URL: (assessed: 20.11.2022).
Amisha, Malik P, Pathania M, Rathaur VK. Overview of artificial intelligence in medicine. J Family Med Prim Care. 2019 Jul;8(7):2328-2331. doi: 10.4103/jfmpc.jfmpc_440_19. PMID: 31463251; PMCID: PMC6691444.
What is machine learning? [Electronic resource]: IBM. URL: (assessed: 20.11.2022).
Deo RC. Machine Learning in Medicine. Circulation. 2015 Nov 17;132(20):1920-30. doi: 10.1161/CIRCULATIONAHA.115.001593. PMID: 26572668; PMCID: PMC5831252.
What is supervised learning? [Electronic resource]: IBM. URL: (assessed: 20.11.2022).
Spyns P. Natural language processing in medicine: an overview. Methods Inf Med. 1996 Dec;35(4-5):285-301. PMID: 9019092.
Lin WC, Chen JS, Chiang MF, Hribar MR. Applications of Artificial Intelligence to Electronic Health Record Data in Ophthalmology. Transl Vis Sci Technol. 2020 Feb 27;9(2):13. doi: 10.1167/tvst.9.2.13. PMID: 32704419; PMCID: PMC7347028.
Lee S, Kim HS. Prospect of Artificial Intelligence Based on Electronic Medical Record. J Lipid Atheroscler. 2021 Sep;10(3):282-290. doi: 10.12997/jla.2021.10.3.282. Epub 2021 Jul 13. PMID: 34621699; PMCID: PMC8473961.
Coorevits P, Sundgren M, Klein GO, Bahr A, Claerhout B, Daniel C, Dugas M, Dupont D, Schmidt A, Singleton P, De Moor G, Kalra D. Electronic health records: new opportunities for clinical research. J Intern Med. 2013 Dec;274(6):547-60. doi: 10.1111/joim.12119. Epub 2013 Oct 18. PMID: 23952476.
Honavar SG. Electronic medical records – The good, the bad and the ugly. Indian J Ophthalmol. 2020 Mar;68(3):417-418. doi: 10.4103/ijo.IJO_278_20. PMID: 32056991; PMCID: PMC7043175.
Hill RG Jr, Sears LM, Melanson SW. 4000 clicks: a productivity analysis of electronic medical records in a community hospital ED. Am J Emerg Med. 2013 Nov;31(11):1591-4. doi: 10.1016/j.ajem.2013.06.028. Epub 2013 Sep 21. PMID: 24060331.
Choudhury S, Chohan A, Dadhwal R, Vakil AP, Franco R, Taweesedt PT. Applications of artificial intelligence in common pulmonary diseases. Artif Intell Med Imaging 2022; 3(1): 1-7. URL: DOI: https://dx.doi.org/10.35711/aimi.v3.i1.1
Seah J, Brady Z, Ewert K, Law M. Artificial intelligence in medical imaging: implications for patient radiation safety. Br J Radiol. 2021 Oct 1;94(1126):20210406. doi: 10.1259/bjr.20210406. Epub 2021 Jun 23. PMID: 33989035; PMCID: PMC9328044.
De A, Sarda A, Gupta S, Das S. Use of Artificial Intelligence in Dermatology. Indian J Dermatol. 2020 Sep-Oct;65(5):352-357. doi: 10.4103/ijd.IJD_418_20. PMID: 33165383; PMCID: PMC7640800.
Diabetic retinopathy [Electronic resource]: Mayo Clinic. URL: (assessed: 20.11.2022).
Grzybowski A, Brona P, Lim G, Ruamviboonsuk P, Tan GSW, Abramoff M, Ting DSW. Artificial intelligence for diabetic retinopathy screening: a review. Eye (Lond). 2020 Mar;34(3):451-460. doi: 10.1038/s41433-019-0566-0. Epub 2019 Sep 5. Erratum in: Eye (Lond). 2019 Dec 10;: PMID: 31488886; PMCID: PMC7055592.
Marinovich ML, Wylie E, Lotter W, Pearce A, Carter SM, Lund H, Waddell A, Kim JG, Pereira GF, Lee CI, Zackrisson S, Brennan M, Houssami N. Artificial intelligence (AI) to enhance breast cancer screening: protocol for population-based cohort study of cancer detection. BMJ Open. 2022 Jan 3;12(1):e054005. doi: 10.1136/bmjopen-2021-054005. PMID: 34980622; PMCID: PMC8724814.
Siontis KC, Noseworthy PA, Attia ZI, Friedman PA. Artificial intelligence-enhanced electrocardiography in cardiovascular disease management. Nat Rev Cardiol. 2021 Jul;18(7):465-478. doi: 10.1038/s41569-020-00503-2. Epub 2021 Feb 1. PMID: 33526938; PMCID: PMC7848866.
Wu M, Lu Y, Yang W, Wong SY. A Study on Arrhythmia via ECG Signal Classification Using the Convolutional Neural Network. Front Comput Neurosci. 2021 Jan 5;14:564015. doi: 10.3389/fncom.2020.564015. PMID: 33469423; PMCID: PMC7813686.
Attia ZI, Harmon DM, Behr ER, Friedman PA. Application of artificial intelligence to the electrocardiogram. Eur Heart J. 2021 Dec 7;42(46):4717-4730. doi: 10.1093/eurheartj/ehab649. PMID: 34534279; PMCID: PMC8500024.
Kim G, Ahn D, Kang M, Park J, Ryu D, Jo Y, Song J, Ryu JS, Choi G, Chung HJ, Kim K, Chung DR, Yoo IY, Huh HJ, Min HS, Lee NY, Park Y. Rapid species identification of pathogenic bacteria from a minute quantity exploiting three-dimensional quantitative phase imaging and artificial neural network. Light Sci Appl. 2022 Jun 23;11(1):190. doi: 10.1038/s41377-022-00881-x. PMID: 35739098; PMCID: PMC9226356.
Yakimovich A. Machine Learning and Artificial Intelligence for the Prediction of Host-Pathogen Interactions: A Viral Case. Infect Drug Resist. 2021 Aug 20;14:3319-3326. doi: 10.2147/IDR.S292743. PMID: 34456575; PMCID: PMC8385421.
Ahmed I, Jeon G. Enabling Artificial Intelligence for Genome Sequence Analysis of COVID-19 and Alike Viruses. Interdiscip Sci. 2022 Jun;14(2):504-519. doi: 10.1007/s12539-021-00465-0. Epub 2021 Aug 6. PMID: 34357528; PMCID: PMC8342660.
Sandeman K, Blom S, Koponen V, Manninen A, Juhila J, Rannikko A, Ropponen T, Mirtti T. AI Model for Prostate Biopsies Predicts Cancer Survival. Diagnostics (Basel). 2022 Apr 20;12(5):1031. doi: 10.3390/diagnostics12051031. PMID: 35626187; PMCID: PMC9139241.
Bodenstedt S, Wagner M, Müller-Stich BP, Weitz J, Speidel S. Artificial Intelligence-Assisted Surgery: Potential and Challenges. Visc Med. 2020 Dec;36(6):450-455. doi: 10.1159/000511351. Epub 2020 Nov 4. PMID: 33447600; PMCID: PMC7768095.
Mofatteh M. Neurosurgery and artificial intelligence. AIMS Neurosci. 2021 Aug 6;8(4):477-495. doi: 10.3934/Neuroscience.2021025. PMID: 34877400; PMCID: PMC8611194.
Du C, Li J, Zhang B, Feng W, Zhang T, Li D. Intraoperative navigation system with a multi-modality fusion of 3D virtual model and laparoscopic real-time images in laparoscopic pancreatic surgery: a preclinical study. BMC Surg. 2022 Apr 11;22(1):139. doi: 10.1186/s12893-022-01585-0. PMID: 35410155; PMCID: PMC9004060.
About da Vinci Systems [Electronic resource]: Intuitive Surgical. URL: (assessed: 20.11.2022).
Doyen S, Dadario NB. 12 Plagues of AI in Healthcare: A Practical Guide to Current Issues With Using Machine Learning in a Medical Context. Front Digit Health. 2022 May 3;4:765406. doi: 10.3389/fdgth.2022.765406. PMID: 35592460; PMCID: PMC9110785.
London AJ. Artificial intelligence in medicine: Overcoming or recapitulating structural challenges to improving patient care? Cell Rep Med. 2022 May 17;3(5):100622. doi: 10.1016/j.xcrm.2022.100622. Epub 2022 Apr 27. PMID: 35584620; PMCID: PMC9133460.
Jiang L, Wu Z, Xu X, Zhan Y, Jin X, Wang L, Qiu Y. Opportunities and challenges of artificial intelligence in the medical field: current application, emerging problems, and problem-solving strategies. J Int Med Res. 2021 Mar;49(3):3000605211000157. doi: 10.1177/03000605211000157. PMID: 33771068; PMCID: PMC8165857.
The Moscow Regional Hospital named after Professor V.N. Rozanov became the first medical organization in the Moscow...
November 14, 2022
On 9 November, the Training and Exhibition Area for Demonstration of Innovations in Medicine at the Almazov...
November 9, 2022