Today, the wave of technology is changing, and medical care is undergoing earth-shaking changes as an important area of ​​people's livelihood. The emergence of artificial intelligence has broken the traditional model of many industries and has continued to move toward intelligence. In order to improve medical service capabilities, medical health builders are also trying to integrate more information technology, and artificial intelligence is an important aspect.
During the CHIMA2018 meeting, the Children's Hospital affiliated to Shanghai Jiaotong University and Weining Health jointly held a media communication meeting, and conducted in-depth exchanges around the theme of “Medical AI-From Creationâ€. Dr. Yu Guangjun, Dean of Children's Hospital of Shanghai Jiaotong University, Director of the Department of Imaging, Children's Hospital of Shanghai Jiaotong University, Yang Xiujun, and Dr. Chen Xu, the head of Weining Health Artificial Intelligence Laboratory attended the event to conduct in-depth exchanges on medical AI related issues.
Promoting medical AI development data is the first layer of protection
The integration of artificial intelligence and medical care is not new. For now, medical imaging is a good entry point in the medical field due to its clear diagnostic rules and mature technology. Yu Guangjun explained in the media communication meeting: "Data quality is an important basic premise for the development of artificial intelligence in the medical field. Take AI bone age detection as an example: its data source must be extracted from the hospital image library. Some data with good quality and at the same time marked by experts. Only using such data as the basis of machine learning can make artificial intelligence diagnosis more accurate." He also said: "In the process of medical big data research, there are also some Application scenarios are not as demanding on data quality, but the quality of data used for machine learning diagnostics must pass."
Leverage AI to empower doctors
Director Yang Xiujun said with deep understanding that major hospitals have been eager to automate bone age assessment with artificial intelligence. Moreover, the diagnosis ideas and processes of bone age assessment are limited to the biomedical model, which is different from other imaging medical services that need to be “first positioned and then qualitatively combined with clinical conclusions†in a multi-modal medical model such as bio-psycho-social, especially suitable for AI. Development, so I developed the idea of ​​AI for bone age detection.
When talking about the original intention of the project, Director Yang Xiujun said that in the past, the bone age assessment based on artificial vision, regardless of the mapping method or the scoring method, was very mechanically time-consuming and subjective, and the reference standards were different. The accuracy of the standard map and the applicability of the population are also quite doubtful, which makes it technically inefficient, and the judgment results between doctors and hospitals are very different.
At the same time, imaging physicians, especially pediatric imaging physicians, have large gaps, heavy individual workload, and a strong desire to liberate from mechanical and heavy bone age imaging. In addition, unlike adult general hospitals, the demand for bone age testing in children's hospitals is very large. Nearly 100 cases of bone age DR readings, matching and bone age matching every working day, and the age of bones map books are frequently updated due to excessive reading and consumption.
Nowadays, we can use the AI ​​to read the film to output the bone age diagnosis report even without 30 seconds, only "read" the bone age is to achieve the sub-second meter, which not only saves a lot of time, but also the average absolute error is only 0.43 years, the diagnostic accuracy rate Up to 98%.
The three major points highly summarize the medical AI "experience record"
At present, the development level of artificial intelligence in the medical field can be described as uneven. Although many medical institutions are exploring the medical AI, many projects have been forced to stop. Facing this situation, Dean Yu Guangjun put forward three major points to provide valuable experience for the majority of medical information construction builders.
1. Grasping the pain of demand
The research and development of artificial intelligence must choose the true pain point of demand as the direction of entry. It is necessary to determine the strong demand in the direction of R&D, instead of complying with some so-called “pseudo-demandâ€.
Yu Guangjun said: "After actual research, we found that from the clinical point of view, the demand for AI technology fusion in bone age detection is very strong. Artificial intelligence bone age detection can effectively improve the diagnostic efficiency, and can effectively avoid the difference in doctor level. Problems such as inaccurate diagnosis results caused by other factors."
Yang Xiujun said that based on artificial vision, the bone age assessment of imaging, regardless of the mapping method or the scoring method, is very mechanical, time-consuming and subjective, and the reference standards are different, the accuracy of the standard map and the applicability of the population are doubtful. As a result, its technical efficiency is low, and the judgment results between doctors and hospitals are very different.
At the same time, imaging physicians, especially pediatric imaging physicians, have large gaps, heavy individual workload, and a strong desire to liberate from mechanical and heavy bone age imaging. In addition, unlike adult general hospitals, children's hospitals have a high demand for bone age detection. Nearly 100 cases of bone age DR readings, matching and bone age matching every working day, and the age-old map books are over-reading and depleting, and need to be updated frequently; In some general hospitals, it is inevitable that the bone age is not accurate enough because of the objective factors such as the small number of subjects and the slow update of the map.
Therefore, both children's hospitals and adult general hospitals are eager to automate bone age assessment with artificial intelligence. Different from other imaging medical services that need to be “first positioned and then qualitatively combined with clinical conclusions†in the multi-modal medical model such as bio-psychological-social, the diagnosis ideas and processes of bone age assessment are limited to biomedical models, especially suitable for AI development. . â€
2. Choose technology maturity
"The image data resources in the medical field are abundant, the imaging diagnosis rules are relatively clear, and most importantly, the current medical image processing technology is also very mature, which provides a good foundation for the development of artificial intelligence in the medical field." The army said.
3. Focus on the ability of partners to engineer
Whether the medical AI project can be successfully landed or not, it is very important to choose a partner. In this regard, Yu Guangjun said: "When selecting partners, we have carefully considered: on the one hand, cooperative institutions should have strong research capabilities; second, cooperative institutions must have strong engineering capabilities. And enrich the practical experience of hospital information construction. At present, we have achieved satisfactory results with Weining Health in artificial intelligence to detect bone age, and improve the efficiency of pediatric imaging doctors while developing the discipline of pediatric imaging and healthy children in China. The establishment of a large sample of bone age database lays the foundation."
Insist on innovation, down-to-earth medical AI "small universe"
AI technology should go deep into the medical essence, solve the core problems in the medical service process, and proceed to empower doctors and benefit patients.
Chen Xu said at the communication meeting: "Weining has two major advantages in the research and development of artificial intelligence. First, we have a large number of hospital customers (more than 5000+ in the country) as the basis. We can firstly meet the various needs of these hospitals. Time perception and feedback have greatly enriched the scene of our artificial intelligence application. Second, we have been focusing on providing medical information complete solutions for medical institutions for many years. The hospital's various data are finally gathered on our data integration platform. This provides a large amount of multi-modal data for research. Weining Artificial Intelligence is committed to seamlessly integrating AI algorithms into hospital business systems. This is not only Weining's innovation, but also the industry's innovation. In the future, we will continue to innovate and down-to-earth to explore truly valuable AI applications."
At the end of the communication meeting, Dean Yu Guangjun said that with the advancement of science and technology, the application scene of medical AI will be more and more abundant. Whether it is in clinical diagnosis, surgical treatment or logistics management, it has broad development prospects and future. Can be expected.
API Powder (Active Pharmaceutical Ingredient)
Active Pharmaceutical Ingredients refer to the raw materials used in the production of various preparations. They are the effective ingredients in the preparations. They are various powders, crystals, extracts, etc., prepared by chemical synthesis, plant extraction or biotechnology, but Substances that the patient cannot take directly.
According to its source,Active Pharmaceutical Ingredients are divided into two categories: synthetic chemical drugs and natural chemical drugs.Chemical synthetic drugs can be divided into inorganic synthetic drugs and organic synthetic drugs. Inorganic synthetic drugs are inorganic compounds (extremely elements), such as aluminum hydroxide and magnesium trisilicate used to treat gastric and duodenal ulcers; organic synthetic drugs are mainly composed of basic organic chemical raw materials, through a series of organic Drugs made by chemical reactions (such as aspirin, chloramphenicol, caffeine, etc.).
Natural chemical drugs can also be divided into biochemical drugs and phytochemical drugs according to their sources.Antibiotics are generally made by microbial fermentation, which belongs to the category of biochemistry.In recent years, many semisynthetic antibiotics are the combination of biosynthesis and chemosynthesis.Among APIs, organic synthetic drugs account for the largest proportion in variety, yield and output value, which is the main pillar of chemical pharmaceutical industry.The quality of API determines the quality of preparation, so its quality standards are very strict. All countries in the world have formulated strict national pharmacopoeia standards and quality control methods for its widely used API
Active Pharmaceutical Ingredient,API powder,minoxidil,monobenzone
PYSON Co. ,Ltd. , https://www.pysonbio.com