News Digest 10.07.2024 — 12.07.2024
Regulators
The Government has approved the deadlines for mandatory labeling of gloves and technical rehabilitation equipment
The Government of the Russian Federation has approved the deadlines for mandatory labeling of five types of technical rehabilitation equipment (TRE) and medical gloves. In the case of crutches, support and tactile canes, orthoses and functional units of prostheses, anti-decubitus pillows and mattresses, urine and ostomy bags, as well as chairs with sanitary equipment, mandatory labeling will start on September 1, 2024, and in the case of medical gloves - on March 1, 2025.
The amendments were made to Government Resolution No. 792 of April 28, 2018. It includes new types of technical rehabilitation equipment, as well as medical gloves. Earlier, the Ministry of Industry and Trade and the Government approved regulations for the labeling of goods and preliminary deadlines for the introduction of mandatory labeling, but they were only added to the consolidated government list now. In October 2023, the Government of the Russian Federation approved an experiment on voluntary marking of technical rehabilitation equipment. The stated goal of the experiment was to combat the illegal import, production and circulation of technical rehabilitation equipment, including counterfeit ones, as well as to increase the collection of taxes and fiscal payments. The pilot project started on October 15, 2023 and is scheduled to be completed in August 2024.
The parameters for marking technical rehabilitation equipment were approved by Government Resolution No. 744 of May 31, 2024. The circulation of products that have not been marked and were put into circulation before the deadline will be allowed until February 28, 2025. Participants in the technical rehabilitation equipment circulation system will transmit information on the withdrawal of technical rehabilitation equipment from circulation from March 1, 2025, and data on their circulation on the market - from October 1, 2025.
A voluntary pilot project for glove marking was launched at the end of June 2024. The decision was made "due to the large share of counterfeit products on the market" and is intended to support domestic manufacturers who are faced with unscrupulous participants selling foreign-made products under the guise of Russian ones.
The marking will affect all main types of gloves - examination, surgical, latex, nitrile, according to eight OKPD 2 codes and 36 codes of the type of medical product in accordance with the nomenclature classification of medical products. A month will be given to assign codes to products imported before March 1, 2025. The circulation of medical products without marking during the transition period is permitted until August 31, 2025. Mandatory marking of medical gloves will start on March 1, 2025.
Market news
Medscan to build medical center in residential area of Omsk for 1.2 billion rubles
A multidisciplinary medical center for children and adults of JSC Medscan will appear in Omsk on the territory of the residential area "Green River", the company's press service told MV. The project will be implemented within the framework of the agreement on development of healthcare infrastructure of the region, signed with the government of Omsk region.
The expected volume of investments in the facility is about 1.2 billion rubles, the implementation period is 2027. The capacity of the new medical center is projected at 130 thousand visits per year. According to the region's governor, Vitaly Khotsenko, the launch of the medical center in the residential area "will be an important step towards social and economic development" of the region.
JSC Medscan is equally owned by Evgeny Tugolukov and JSC Rosatom Health Technologies (formerly Rusatom Healthcare). The company is represented in 105 cities in 31 regions of Russia. The group includes 65 medical centers, 15 laboratories and 431 laboratory offices under the KDL brand, as well as diagnostic centers, hospitals with a surgical inpatient department, multidisciplinary and specialized clinics (including oncology clinics), and research and development centers.
Based on the results of last year, JSC Medscan reported a record growth in revenue - it increased fourfold, to RUB 22.4 billion. EBITDA (the company's earnings before interest, taxes, depreciation and amortization) increased 12 times. The revenue of the Medicine business line amounted to RUB 9.7 billion, an increase of 40% compared to 2022. The revenue of the Laboratories business line reached RUB 12.8 billion.
Invitro plans to launch its own network of clinics
The Invitro Group of Companies, which specializes in laboratory diagnostics, intends to strengthen its medical direction - for this purpose, it plans to buy out one of the private federal networks of clinics or a group of regional networks, the company's CEO Yuri Leonov told Vedomosti. At the same time, Invitro may not stop at buying one network, he said. "We are doing pilots within the company to understand possible formats for converting laboratory traffic into clinics, and we are considering ideas on how to enter this market through M&A," Leonov explained. He did not specify the amount of funding allocated for these purposes. In June, he said that Invitro was restructuring its development policy with an emphasis on opening its own medical offices, MV reported. Currently, about 50% of the company's medical offices are franchises. Current franchisees wishing to exit the business will be offered a buyback. Invitro is ready to spend about 2 billion rubles in the next two to three years on this, as well as on upgrading its medical offices.
According to Igor Sulkis, an M&A specialist and former president of Medscan Group, there are many opportunities for synergy between medical and laboratory institutions. He noted that attracting a patient from a laboratory to a clinic will require less expense.
Source: https://medvestnik.ru/content/news/Invitro-planiruet-zapustit-sobstvennuu-set-klinik.html
Other
Study: AI copes with diagnosis worse than doctors
A group of scientists from Germany, Great Britain and the USA published a study in Nature Medicine on the accuracy of diagnosis by large language models (LLM) and the possibility of introducing such technologies into the healthcare system. The authors tested this type of artificial intelligence (AI) on a database containing 2.4 thousand cases of patients with four common abdominal pathologies (appendicitis, pancreatitis, cholecystitis and diverticulitis). The researchers concluded that LLM diagnoses patients with low accuracy. Scientists also point out the difficulty of introducing these technologies into real work processes due to frequent violations of LLM instructions, due to their sensitivity to the amount and order of information provided for analysis. The authors of the publication explain that the LLMs, which include Llama 2, developed by Meta (recognized as an extremist organization in the Russian Federation and banned), medical language models Clinical Came and Meditron, were tasked with making a diagnosis and creating a treatment plan based on the patient's medical history (results of physical examinations, lab tests, and imaging reports). The researchers checked not only the accuracy of the AI's verdict, but also its compliance with diagnostic and treatment recommendations, the accuracy of interpreting tests, and the ability of the language models to follow instructions. The scientists compared the behavior and diagnosis of LLMs with the decisions of practicing doctors. In terms of the accuracy of diagnosis of 80 randomly selected patients, as the study shows, LLMs are significantly inferior to clinicians. The AI made the greatest number of errors when determining cholecystitis and diverticulitis: the accuracy of the diagnosis made by the language models in these cases ranged from 13% to 68%, while doctors determined these diseases in 84–86% of cases. The AI was able to compete with the accuracy of doctors' conclusions only in detecting appendicitis - the accuracy of the diagnosis of some language models exceeded the indicators demonstrated by doctors by 3-4%. The authors of the study explained that LLMs can qualitatively determine some pathologies, but do not currently have the "diagnostic range" of practicing doctors.
In the next experiment, which tested only Llama 2 and its derivative language models (since medical LLMs are not designed to perform tasks "not embedded" in them), the researchers asked the AI to independently indicate all the information it needed to diagnose a patient. Under these conditions, the average accuracy of the diagnosis made by language models decreased by another 13% and amounted to 45-55%.
During the second experiment, specialists found that LLMs often do not prescribe examinations prescribed by diagnostic recommendations, do not follow treatment recommendations, especially in cases with more severe forms of pathologies, and are unable to interpret the results of laboratory tests. While testing the ability of LLM to follow instructions and generate responses that could be quickly used in other departments of the clinics, the scientists found that the AI made errors in its algorithms for every second to fourth patient, and also suggested using non-existent tools in every second to fifth case.
In addition, the responses of the language models differed with minor changes in the wording of queries. In particular, the AI gave different answers when the query for “final diagnosis” was changed to “primary diagnosis” or “main diagnosis”. The accuracy of the diagnosis made by LLM, the study reports, also decreases when providing language models with access to a larger number of test results. At the same time, for different pathologies, the accuracy of the AI's conclusions reaches the highest levels when providing a specific list of medical studies.
The analysis was conducted by employees of the Technical University of Munich, the Helmholtz Center and the University Hospital of the Ludwig Maximilian University of Munich in Germany, as well as representatives of Imperial College London in the UK and a network of private non-profit hospitals in the USA. The researchers emphasized that the potential of LLM in healthcare had previously been tested in structurally simple medical contexts, where AI results often exceeded those of practicing doctors.
The danger of uncontrolled use of AI in healthcare was also noted by the WHO in July 2024, in an updated version of the laboratory biosafety guidelines. The document emphasized that AI can help improve public health, but this tool can also be used for other purposes, in particular, to create new viruses and toxic materials. In addition, the WHO noted that models can generate unreliable information due to the data they were trained on.
Newly registered medical devices
We publish a list of new medical devices registered from 07/05/2024 to 07/12/2024
Link to the list: [see. table in attachment]
News
