Artificial intelligence in Moscow begins to issue conclusions on fluorography without the participation of a doctor

2024: Artificial intelligence in Moscow begins to issue conclusions on fluorography without the participation of a doctor

Artificial intelligence in Moscow begins to issue conclusions on fluorography without the participation of a doctor. Metropolitan Mayor Sergei Sobyanin wrote about this in his Telegram channel on April 24, 2024.

In Moscow polyclinics, artificial intelligence will analyze X-rays without the participation of a doctor. An average of two million fluorographs and chest radiographs are performed annually, and most of them are carried out for prevention. For example, 99% of fluorographies do not detect pathologies, and doctors are engaged in the description of the norm. Given the vast accumulated experience in this area, smart algorithms are able to accurately determine the absence of signs of diseases. Therefore, we are ready to convey the analysis and description of studies configured for maximum sensitivity to AI services. For this, a special compulsory medical insurance tariff has been introduced, - said Sobyanin.

mos.ru

If the results of the studies do not reveal pathologies, then the conclusions of the AI will automatically go to the electronic medical records of patients.

To guarantee the absence of errors, until September 2024, the results of the AI analysis will be rechecked by radiologists. At the first stage, the work of AI models will be supervised by specialists from the clinic named after Professor Yu.N. Kasatkin of the Russian Medical Academy of Continuing Professional Education. Its employees will analyze and prepare a conclusion within 24 hours. The Moscow mayor's office added that all services will be free.

The mayor's office believes that the use of artificial intelligence will increase the quality of research and at the same time free doctors from routine work to analyze the same type of research without pathologies.^[1]

2023: Moscow City Hall told how an AI tool works to make a final diagnosis of Aida

In September 2023, the Moscow mayor's office talked about the creation of an AI tool to help doctors in the capital's polyclinics. The diagnostic assistant "Aida," developed jointly with "Sber," helps therapists and general practitioners in making the main final diagnosis based on the data of the electronic medical record (EMR) over the past two years. Read more here.

2020

500 doctors from 300 medical institutions joined the experiment on the use of AI in Moscow

As part of an experiment on the use of artificial intelligence (AI) in medical practice, these technologies are used by almost 300 medical institutions to diagnose diseases using X-ray, the Moscow government portal said on November 16, citing Deputy Mayor for Social Development Anastasia Rakova^[2].

More than 500 radiologists are involved in the experiment. They use 38 digital AI services that 21 companies have developed. The DIT of Moscow, participating in the implementation of the experiment, could not promptly give TAdviser examples of the IT products involved in the experiment and clarify the fate of the project for the development of a system based on open technologies, which was discussed in 2018 (see the block below).

Anastasia Rakova says that it is planned to make the use of AI technologies the norm for urban health care "(photo - M. Mishina, press service of the mayor and the government of Moscow)"

A year ago, we began an experiment to introduce artificial intelligence into the capital's health care system. Our goal is to increase the accuracy of diagnostics and reduce the burden on staff. We went on a model of open innovation. That is, we ourselves do not develop computer vision systems, but involve developers, work out scenarios for the application of these systems in clinical practice, integrate these solutions with our ERIS EMIAS service . Next, we check the services for accuracy and the necessary functionality using a special method, select the best options for introducing a doctor into the work, - the portal of the Moscow government quotes Rakova.

According to the vice mayor, on average, about 4.5 thousand instrumental studies per day are processed using such technologies.

In medical institutions, AI, integrated into the EMIAS service, analyzes the results of several types of radiation studies: computed tomography, X-ray diagnostics, mammography and fluorography. They are used to diagnose pneumonia, lung cancer and other pulmonary pathologies, breast cancer, as well as to detect COVID-19.

According to Sergei Morozov, director of the Center for Diagnostics and Telemedicine, doctors work with both conventional X-ray images and processed AI. The pictures arrive at the base a few minutes after they have been taken. The base includes millions of images and is constantly being replenished.

The experiment will last until 2021. Anastasia Rakova says it is planned to make the use of AI technologies the norm for urban health care.

The beginning of the urban experiment

According to Moscow Mayor Sergei Sobyanin, an experiment on the use of artificial intelligence (AI) technologies to analyze medical images was launched in February 2020. He wrote about this on his page on the social network "VKontakte."

The experiment on the use of innovative technologies in the field of computer vision for the analysis of medical images is being implemented by the Moscow Department of Health in conjunction with the DIT of Moscow. It covers the use of computer vision technology for recognizing signs of cancer and other diseases in medical images, said earlier Moscow Vice Mayor Anastasia Rakova. It is expected that it will reduce the likelihood of errors and risks for patients of Moscow medical institutions.

According to Rakova, to assess the possibility of introducing such technologies, a phased analytical approach will be required. In this matter, the expert opinion of not only doctors, but also specialists developing such technologies is important.

Therefore, for the experiment, we consider it right to attract companies working with computer vision, - said Rakova.

During the experiment, the program will analyze 3 million X-ray and tomographic images, which are contained in the database of the Unified Radiological Information Service (ERIS). The system will learn to find signs of pathology in these images, and then determine them in the newly received images.

In the future, according to the idea, the program should automatically track changes in the pictures of the same patient, and give an opinion on the dynamics to help doctors make the right decisions when making a diagnosis.

2018: Moscow creates an artificial intelligence system for doctors on open source software. IBM solutions did not fit

Moscow is developing its own system based on artificial intelligence technologies and big data for use in healthcare, told TAdviser in the city's Department of Information Technology (DIT) in February 2018. It is designed to support doctors in making medical decisions.

At the first stage, the development is carried out in two directions: the recognition of lung cancer on CT studies, as well as the recognition of heart diseases on ECG studies. These directions were chosen because the most common cause of death is diseases in the field of oncology and cardiology, TAdviser explained in DIT.

The solution is based on the known architectures of neural networks. At the heart is an open software library for machine learning developed by Google - TensorFlow.

Before the development of its own system, Moscow tested IBM solutions for healthcare, but they were not suitable for use in Russian practice

DIT works with other areas of medicine, the knowledge base accumulates. To train neural networks, specialists are engaged in marking diseases in their research. To speed up the processing of large amounts of medical data, DataSet's own preprocessing library was developed.

At the initial stage, the DIT analytical center is developing neural network algorithms to prove the accuracy of recognition and business application, the department told TAdviser. The next stage will test and publish the results.

After that, a competition can be announced for the refinement of software based on a trained neural network or the introduction of a disease recognition module at EMIAS. But it is too early to talk about this, the DIT said.

The final accuracy assessment is scheduled to be done in the summer of 2018. The appropriate estimator is currently being selected. Perhaps testing will be carried out at the site of the Moscow Center for Radiology (NPTSMR) in partnership with which algorithms are being developed, - says the representative of DIT.

Earlier, in January, Moscow Mayor Sergei Sobyanin at a meeting with doctors - winners of the Leadership. Med competition said that IBM solutions were tested in the city to support medical decision-making, but they could not be shifted to Russian practice.

IBM offers 2 products in healthcare. This is IBM Watson - to support medical decisions regarding prescriptions, as well as IBM Oncology Expert Advisor - for cancer diagnosis, aimed mainly at lung cancer.

DIT says that "these solutions did not go into full-fledged products, and remained expensive research developments."

The department explained to TAdviser that the use of IBM solutions in Moscow was considered inappropriate for two reasons. First, they do not have a Russian-language version. Secondly, they are not adapted to the current electronic map: IBM Watson is trained on disparate and diverse data of US electronic maps.

The main problem of all such solutions existing on the market today is the lack of evidence of metrics of accuracy, effectiveness, sensitivity to disease recognition. All such systems work on closed data and on closed program code, - explained TAdviser in DIT.

IBM's Russian office could not comment on TAdviser's project to test its solutions in Moscow healthcare.

In 2018, DIT representatives will participate in the MICCAI (Medical Image Computing and Computer Assisted Intervention) conference. During the event, they will try to prove the accuracy of developments based on open data, the department added.

Sergei Sobyanin previously estimated that Moscow will have the largest amount of data from electronic medical records. According to him, as of January 2018, 6.5 million Muscovites have such cards.

According to the mayor, in the field of introducing artificial intelligence into medical activities, Moscow has a lag behind Western countries. At the same time, in general, the city does not lag behind in organizing information support for medicine, he believes.

Notes