Where is the breakthrough point of personal intelligence? The situation of medical fragmentation needs to be standardized, and a group of emerging technology companies stand out.
Among them, this year's AI+ medical field financing leader, harvesting 300 million US dollars, the emergence of Yitu. 2018 is a milestone year for Etu Medical , which became the first unicorn company in the field of medical big data and artificial intelligence.
From its inception in 2009 to its current independence, Etu Medical has adopted a full-stack development approach that not only applies medical artificial intelligence technology to the diagnosis of various diseases. The use of the most local data structure to build a knowledge map, the use of algorithms and big data applied to the analysis of cases and literature to obtain its greatest value.
In the tail of 2018, Arterial Network had the honor to invite Dr. Fang Wei, the vice president of Yitu Medical, to come to the forum and gave a keynote speech on "The future of AI is medical".
Dr. Fang Wei, Vice President of Yitu Medical, gave a keynote speech on "The future of AI is medical"
Advanced AI capabilities will occur in the medical field
Why must the advancement of AI capabilities occur in the medical field? Dr. Fang Wei believes that the richness and complexity of medical clinical scenes give AI the ability to advance. What is this path? We can divide it into three phases.
When AI research started, companies focused on single-task-centric medical solutions. There are more than 100 AI medical startups across the country, 90% of which focus on the detection of pulmonary nodules. This is because lung nodules have open data sets, and companies can use this as a breakthrough to gradually expand into other diseases.
In practice, the steps for patients to seek medical treatment include: 1. Regular examination (predicting how the patient performs the next examination through the main complaint, etc.); Second, imaging diagnosis; Third, MDT (how to design a treatment plan for the patient). In contrast, the current stage of image AI plays a moderate role in the entire process, but there is still room for expansion.
Single-agent AI, such as pulmonary nodules, can identify pulmonary nodules, and macular degeneration can recognize macular degeneration, but if the patient's lungs have lung water, lung patches, cystic lesions, single-task AI is Powerless. Is there a need for a single disease in a hospital department? The answer is yes. But is the demand large enough to support a complete department operation? Far from enough.
Etu Medical is at the forefront in this respect, jumping out of the thinking of single-agent diseases. Nowadays, according to the care of the map. Ai ? The chest CT intelligent 4D imaging system has been able to identify most lesions in the lungs based on a set of images, providing doctors with intelligent treatment advice, similar case recommendations, and automatic historical image comparison.
The second phase of medical AI is a multidisciplinary diagnosis and treatment, that is, a site-centered medical solution. According to the doctor's thinking, a lung CT can often understand the nodules, patches, and effusions in the patient's lungs, and make a diagnosis of the patient's chest based on comprehensive judgment. At the same time, doctors can also propose a comprehensive lung cancer solution based on the patient's genetic test results, combined with his previous medical history, complaints, and current medical history. This is a relatively high-level stage of human medicine, but this is not the limit of future medical care.
Patient-centered medical solutions are the ultimate goal of future healthcare and the third growth path for AI in healthcare. Patient-centered means that every natural person begins to be born and all information about his body is electronic. This scene is not far away from us. In practice, some countries in northern Europe, such as Iceland, have a closed-loop population and a small population base. It is relatively difficult to realize lifelong informationization of citizen information. So every patient, or every citizen, from birth to death, has a full link of medical information in the information center.
What are the two core elements for achieving this path advancement? Dr. Fang Wei believes that the problem can be solved from two directions: one is
The technical strength of human beings; the second is the ability to use this powerful technology to unlock clinical application scenarios.
In general, it is the establishment of a task-centered intelligent department, a smart diagnosis and treatment centered on diseases or parts, and a patient-centered intelligent hospital.
In this regard, Etu Medical has chosen three technical directions:
1. Image-based visual recognition, including information from the imaging department, ultrasound data, and pathological data.
2, natural language processing technology, a large amount of medical data exists in the way of medical history natural language, how to connect these huge data, explore the connections, and try to emerge from some unexpected creation, which is the informationization of medical data significance.
3, voice platform, doctors often need two sensory abilities to understand information - vision and hearing. As visual technology has matured, auditory technology is still being developed. This is because sound information is more vague and more ambiguous than image information, and is more susceptible to the subjective consciousness of the utterer himself. To understand the meaning of sound information, you need to understand not only the statement itself, but also the environment in which the subject is located. In the future ward conceived by mankind, sound information can even replace input abilities, complete texts with sound, and apply to a wider range of fields.
The research and development team of Etu Medical is composed of top university graduates such as MIT and Tsinghua Peking University. It has worked for leading Internet companies such as Google and BAT. This means that Etu Medical has excellent entrepreneurial and research and development genes. Map Medical has won various world-class awards.
Today, Etu Medical has recruited 400+ senior physicians to form a team, full-time director of the top three hospitals to participate in full-time guidance. Collaborate with Internet talent to guide the products and services of Etu Medical.
In terms of data annotation, Vision Medical has invested a large amount of funds and capabilities to ensure the correctness and availability of standard data. This is an insurmountable barrier based on the accumulation of time and manpower.
In this context, Etu Medical has gradually developed into the only company in China that can cover full-link medical data and provide full-stack clinical medical products.
In less than two years, Yitu Medical's products have been embedded in the clinical work field, and more than 200 top three hospitals have deployed the products of Yitu. Among the top 100 hospitals in China, more than 50 hospitals are equipped with products from Yitu. Take care. Ai? Children's growth and development intelligent diagnosis system, for example, this product combines physiological indicators to evaluate the comprehensive growth and development of children, and gives a customizable growth and development report, which brings great convenience for clinical diagnosis, teaching and scientific research. Eight of the top 10 children's hospitals in the country have already used the product for clinical use, and the top hospital embedding rate is as high as 80%.
In the process of debugging product performance and improving product experience, Yitu Medical not only stays sensitive and specific. This is because sensitivity and specificity are strongly dependent on the data set of the input algorithm. For companies and third parties outside the hospital, we can't judge the source of sensitivity and specificity of their AI products. It is a cleaned data set or a data set for actual clinical use.
Therefore, the KPI used by Etu Medical for team or product evaluation is called the clinical report adoption rate. Nowadays, the clinical report adoption rate of some mature products according to the map has exceeded 92%, that is, for every 100 reports of AI diagnosis, 92 doctors have no problem to review and report directly to patients. In other words, the Yitu Medical AI product has achieved very high recognition in clinical use.
As one of the four largest AI unicorns in China, the scale of the company has far exceeded that of pure AI artificial intelligence startups. Recently, the business of Etu has covered the whole world, and its coverage city has extended to the United States, Europe, Southeast Asia, Africa... In terms of cooperation, Etu has started global business expansion with strategic partners Huawei and Microsoft, and faster deployment is underway. In progress...
Smart Home,Mini Gps Signal Tracker,Smart Door Window Sensor
Shenzhen Bio Technology Co., Ltd , https://www.huifantech.com