Consultation and registration, consultation and medicine, surgical rehabilitation...... The medical process, which seems to be the same as before, has quietly changed before many people are aware of it, and behind these changes is the application of artificial intelligence in the medical field. In the healthcare sector, the rise of new quality productivity means innovation-driven efforts to improve the quality and efficiency of healthcare services. A reporter from Beijing Youth Daily learned that at the annual meeting of the Zhongguancun Forum and related supporting activities, many cases in the medical field based on big data models and artificial intelligence empowerment have made people see a corner of the future of medical care, and these achievements have been applied in our daily life.
Intelligent triage shortens the time spent in the hospital
If you feel that you have a headache and need to go to the doctor, should you hang the number of gods or gods, or should you go to the clinic for pain? Many people have encountered such a scene when they seek medical treatment: they feel uncomfortable, but they can't tell what is uncomfortable. In today's situation of subdivision of hospital departments, it is even more unclear which department number should be hung. There are also many patients who have been listed in a certain department, but after the doctor has asked once, they said that they should be listed in another department.
"Our hospital-specific services address the confusion patients face when registering. At the Zhongguancun International Technology Trade Conference of the annual meeting of this Zhongguancun Forum, the staff of Yidu Technology is introducing to the consultants. In the past 10 years, it has been authorized to process and analyze more than 4 billion medical records and accumulate a large number of multi-dimensional quantifiable knowledge graphs. Through self-developed data generation technology, these knowledge graphs are used for large model training.
"Many patients don't know how to describe their condition. Our app is more like a physician assistant than an existing AI conversation. It has rich medical knowledge, and is familiar with the configuration of hospital departments and the diseases that each department is good at diagnosis and treatment, and can provide patients with accurate department recommendations through the analysis and matching of patients' conditions. ”
In the app test, one of the testers was suspected of sleep-related leg cramps after answering several symptom questions about the AI and recommended that they seek medical attention as soon as possible. The staff member also said that the service can also provide a pre-consultation function, simulating the dialogue between doctors and patients, understanding the patient's physical condition and symptoms, generating pre-consultation medical records, and improving the efficiency of diagnosis and treatment and patient experience.
Some hospitals in Beijing have started to use smart systems based on big data to improve the patient experience. Zhao Wei, vice president of Fuwai Hospital of the Chinese Academy of Medical Sciences, said at the medical AI industry integration and development conference of the annual meeting of the Zhongguancun Forum that the application of intelligent triage in the hospital not only accurately recommends and matches special doctors for first-time patients, but also directly recommends doctors in previous outpatient visits or ward team doctors for returning patients.
The adoption rate of intelligent triage in the hospital has reached 78.2%, and the accuracy rate of specialty triage has reached 97.4%.
At present, the system used by Fuwai Hospital can not only realize intelligent triage before registration, but also save time for patients in the hospital consultation, examination, medicine and other links. Zhao Wei revealed that after using the system, the average waiting time for in-hospital examination of patients has been reduced by 21 minutes compared with 2019, and the overall saving from admission to discharge has been 40 minutes.
AI-assisted systems make diagnosis and treatment more accurate
A reporter from Beiqing Daily learned that compared with being able to see a doctor faster, for patients, the medical changes brought about by big data models and artificial intelligence occur in places they can't see, and more doctors feel the changes brought by these technologies to diagnosis and treatment.
Taking the most basic electrocardiogram of cardiac examination as an example, the ECG assisted decision support system with independent intellectual property rights of Fuwai Hospital covers 48 kinds of ECG diagnosis on the basis of a large model. "There are more than 140 interpretations and diagnoses on ECG, and 98% of people only use 4 of them. Doctors with different experience also have inconsistent ECG interpretation results. This brings certain problems to the later diagnosis and treatment. Zhao Wei said that the accuracy rate of the ECG assisted decision support system is very high, and the effect is equivalent to the level of 5 to 10 years of senior ECG doctors. On average, 93% of the diagnostic prompts of the system will be accepted in clinical practice, which shortens the time for doctors to interpret by 63.3%.
It not only improves the efficiency of doctors, but also saves manpower. "The number of ECG readings in Fuwai Hospital has reached 1,700 to 2,000 a day, and it takes about 7 or 8 senior doctors to interpret them. Zhao Wei said that with this system now, the final interpretation results are reviewed by high-level ECG doctors. After the whole process, only 2 to 3 ECG doctors are needed.
Not only the interpretation of basic examinations such as electrocardiogram and blood pressure, but also some diseases that require expert-level experience to be judged, now with the assistance of big data models and artificial intelligence.
"For medical resources, the expert experience of a tertiary hospital is very important, for example, the same pulmonary nodules can be found by experts by looking at the film, while inexperienced doctors may miss the diagnosis. Guo Xiaolong, public relations director of Lingkr Technology, said that the company has completed the in-depth learning and training of tens of thousands of patient data in about two years with the help of cooperative hospitals, and the artificial intelligence-assisted diagnosis software products are improving day by day.
"This product can assist doctors to quickly and accurately find lung nodule lesions, and analyze them for benign and malignant analysis, realize early screening and early diagnosis of lung cancer, and reduce the rate of missed diagnosis and misdiagnosis of lung cancer. Guo Xiaolong said that now its products have been well used in outpatient clinics, imaging departments, and thoracic surgery, and have built the first artificial intelligence-assisted diagnosis platform for pulmonary nodules in Beijing, Tianjin and Hebei.
Even for surgery, AI can make surgery less risky and patients recover faster.
"Before the surgery, our system tailors a preoperative plan based on information such as the patient's medical records and images. For surgeons, as long as they follow our pre-operative protocol, the surgery can achieve a more accurate result. Yu Haiying, general manager of Beijing Natong Intelligent Technology Co., Ltd., said.
The reporter of Beiqing Daily learned that Renato's "knee replacement surgery navigation and positioning system" obtained the registration certificate of Class III medical devices of the State Food and Drug Administration last year, as the first domestic knee unicompartmental replacement surgical robot, filling the technical gap in the field of domestic robot unicompartmental replacement.
Yu Haiying said that the preoperative plan means that before the patient enters the operating room, the consumables to be adapted, the model of the implant, and the angle and depth of the osteotomy or grinding can be planned in advance, so that the doctor can have a systematic understanding of the patient before the operation. The standardized operation process of the system also makes the human-computer interaction more comfortable, which not only effectively shortens the operation time, but also reduces the amount of intraoperative bleeding and shortens the recovery period because of the precise operation. Compared with the traditional surgery group, the alignment of the affected limb after surgery was closer to the ideal state, which improved the long-term prognosis of the patients.
With the assistance of AI, this kind of surgical equipment that allows doctors to "see" more clearly, "find" lesions more accurately, and operate more stably and more delicately during surgery with the assistance of AI was unveiled at the annual meeting of this Zhongguancun Forum.
In addition to the above-mentioned "Knee Replacement Surgery Navigation and Positioning System", there is also the "Taikon Space DSA" that can simultaneously eliminate murmurs and see the walls of blood vessels during angiography. The product is an AI image processing interventional diagnosis and treatment system, equipped with a new generation of WE-Aimage intelligent supercomputing high-definition image platform, which greatly improves the image quality.
RM-50
Also unveiled was the ROIDMI neurosurgery robot RM-50. This is an AI intelligent segmentation and planning algorithm, a self-developed six-degree-of-freedom robotic arm with self-developed core components, and a multi-spectral tracking and positioning platform, which is suitable for clinical scenarios such as intracerebral hemorrhage aspiration flow, intracranial biopsy, and navigation. "The golden treatment time for cerebral hemorrhage is 6 hours, but in some grassroots hospitals, many doctors suffer from lack of experience and equipment, so they can only refer patients to more advanced hospitals, which is easy to delay the treatment time. According to the on-site staff, this set of equipment can find the location of plaques through AI, accurately locate the point of cerebral hemorrhage, and assist doctors to complete the operation. With such equipment, primary hospitals can also carry out timely treatment of patients with cerebral hemorrhage.
In this exhibition, the most interesting thing is the laparoscopic single-port surgical system that can "hand-peel" raw quail eggshells. Using a glove-like device, the field worker was able to operate a robotic arm at the far end to peel open the shell of a raw quail egg without damaging the soft membrane that was close to the inner wall of the eggshell. "We have more than 600 patents for this system, which relies on artificial intelligence and algorithms to allow the device to filter out the influence of the natural physiological shaking of the doctor's fingers during operation, or some bending of the front end manipulating arm, which increases the stability of the device during the operation. ”
Sharp laparoscopic single-port surgical system
In Yu Haiying's view, the help that artificial intelligence brings to modern surgery is to be more accurate and safer. Not only that, with the assistance of artificial intelligence, it can also effectively shorten the learning curve of young doctors.
Helping young doctors improve the level of diagnosis and treatment as soon as possible and empowering grassroots diagnosis and treatment has become the most widely used big data model and artificial intelligence + medical care. According to data, as of December 1, 2023, 122 medical intelligent software in China have been approved to enter the market, of which 80% are image assistance. With the support of these auxiliary diagnosis and treatment intelligent systems, some diseases can be screened out in the community diagnosis and treatment process, and chronic diseases such as hypertension can be better controlled.
"Prevention of disease" big data assists in early warning
Traditionally, Chinese people believe that "treatment of diseases before they occur" is the priority, and prevention before diseases, timely intervention, and early diagnosis and early treatment have become the consensus in the current medical field. So, how to predict and how to prevent diseases before they occur? Artificial intelligence and big data models have tools for the "early warning" of diseases.
Nowadays, the disease control department reminds the public to pay attention to the fact that a certain type of disease is about to enter a high incidence period, and the news of paying attention to personal protection has been as deeply rooted in the hearts of the people as the weather forecast. Behind this, it is also inseparable from artificial intelligence to empower medical care. "The intelligent decision-making platform developed by us using artificial intelligence technology can help disease control and other departments more accurately predict the development trend and high incidence period of diseases by analyzing the information and information of various diseases in the international health sector, so as to take corresponding prevention and control measures in advance. The staff of Yidu Technology said.
Not only macro trend prediction, but even if it is specific to individuals, artificial intelligence also has a certain ability to predict diseases.
There is an important part of our body that is the "observation post" of disease, which can indicate whether we have diabetes, stroke, high blood pressure and other problems and the degree of development of chronic diseases, which is the eye.
"The eye is the only organ in the body where blood vessels and nerves can be directly observed from the outside. From these blood vessels and nerves, we can determine whether there is vascular disease or neurodegeneration, and infer whether there are some underlying diseases in the body. Wei Wenbin, vice president of Beijing Tongren Hospital, said in a speech at the Medical AI Industry Integration and Development Conference.
This idea was made possible by the quantitative analysis of ocular blood vessels using image analysis techniques.
Artificial intelligence, on the other hand, can learn subtle information that ophthalmologists cannot recognize, and finally complete the task of disease detection with clear labels through the learning and training of retinal changes in patients with a certain disease through big data models.
Wei Wenbin said that some diseases, such as cognitive dysfunction, may have neurobiological changes in the brain 20 years before the onset of clinical symptoms. However, the current drug treatment effect of this disease is not good because the patient may have missed the best time for treatment at the time of onset. Research based on quantitative retinal vascular analysis may be able to revolutionize the early diagnosis of this disease in the future.
Today, optic nerve and retinal endology provides non-invasive images of central nervous system tissue, providing doctors with a window into a patient's neuropathy. Similarly, the geometry of the retinal blood vessels allows doctors to gain insight into changes in other organs, such as the heart and kidneys. Several related academic research papers have been published at home and abroad.
As early as 2020, Mainland China has approved the launch of the "Fundus Image Assisted Diagnosis Software for Diabetic Retinopathy". At present, the software has been applied to the early detection of diabetes in domestic primary diagnosis and treatment units and physical examination institutions.
Artificial intelligence has accelerated the layout of the medical field
According to agency data, the scale of China's digital health market has reached 195.4 billion yuan in 2022, with an average annual compound growth rate of 30.73% in the past five years, and the market size is expected to further increase to 413 billion yuan in 2024.
Today, the medical field is one of the most widely used and effective fields of AI technology. The global investment in artificial intelligence is nearly 10 billion US dollars, of which medical artificial intelligence investment ranks first.
The combination of AI and medical care not only greatly reduces the tedious basic work of medical treatment and improves the accuracy of clinical diagnosis, but also helps to achieve a balanced distribution of medical resources and ensure that different regions can obtain high-quality diagnosis and treatment services.
Even in the near future, with the assistance of AI, the emergence of innovative drugs will accelerate, and the R&D cycle of new drugs will be greatly shortened. Yan Ning, founding dean of the Shenzhen Academy of Medical Sciences, director of the Shenzhen Bay Laboratory, and chair professor of Tsinghua University, mentioned at the annual meeting of the Zhongguancun Forum that the use of AI empowerment and the use of a huge structural library for drug design is no longer empty talk.
The mainland is also accelerating the layout of the artificial intelligence industry in the medical field. In 2019, the National Medical Products Administration officially established the AI medical device standardization technology focal point, and organized the formulation and revision of the mainland AI medical device standards to provide technical support for the supervision and industry.
In October 2021, the Ministry of Industry and Information Technology and the State Medical Products Administration organized the unveiling of artificial intelligence medical device innovation tasks, focusing on 8 categories of unveiling tasks, including intelligent auxiliary diagnostic products, intelligent auxiliary treatment products, and medical artificial intelligence databases, focusing on the two directions of intelligent products and supporting environment, and soliciting and selecting a group of units with strong innovation capabilities to focus on tackling key problems, promoting the innovation and development of artificial intelligence medical devices, and accelerating the application of new technologies and new products.
However, AI is not magic, and at the current stage, AI cannot solve all medical problems. Many experts and scholars believe that, for example, in the research and development of new drugs, the greater role of artificial intelligence is to reduce costs and increase efficiency. The path of AI new drug research and development is to model AI autonomous learning algorithms by obtaining big data, optimize models, machine deep learning, and finally evaluate model performance. These can greatly reduce the time and cost of drug development. However, the AI pharmaceutical model also needs to pass clinical trials, and it is not predicted to be effective.
In fact, last year, six artificial intelligence medical devices in mainland China were finally approved through the fast approval process of the green channel for innovative medical devices. But 2 of these 6 innovative products entered the program back in 2019 and 3 in 2021. This also represents the cautious attitude of the regulatory authorities towards such innovative medical devices.
Zhao Wei, vice president of Fuwai Hospital of the Chinese Academy of Medical Sciences, said in his speech that the premise of the application of medical artificial intelligence is to deal with four elements - data is fundamental, talent is the guarantee, verification is the key, and the most important thing is that security is the bottom line. Just like a pair of calligraphy in the exhibition hall of Yidu Technology - "What can really change medical treatment must be compassion, its eyes must be awe, and its path must be time." ”
Text/Beijing Youth Daily reporter Zhang Xin
Photographer/Beijing Youth Daily reporter Zhang Xin
Edit/Field