论文部分内容阅读
There was a long queue at the diagnostic laboratory in Shanghai’s famed art and tourist destination West Bund Art Center. But strangely， the people in the line did not look tense and anxious as they are wont to when awaiting a diagnostic test. The reason for this became clear when one looked harder. The Future Diagnostic Lab was a demonstration of what labs would be like years away， a display booth at the World Artifi cial Intelligence Conference 2018.
The booth put up by Proxima， a startup providing medical imaging by using artifi cial intelligence （AI）， demonstrated hi-tech virtual diagnosis where visitors lined up for a roleplaying game， being a doctor of the future. By simply clicking on an operation panel， these “doctors” could watch on a big screen how AI technologies facilitated medical examinations， radiography and diagnosis， even coming up with therapeutic suggestions， and all of this within seconds.
The lab displayed the virtual intelligent diagnosing of complicated medical problems like lung cancer. Once a “doctor” chose the lung module on the panel， a three-dimensional scan of the patient’s lungs appeared on screen. A red circle immediately marked out the nodules where the cancerous cells were， a hi-tech method that can save doctors valuable time， making an early diagnosis and pinpointing the areas where the disease has spread.
“With the assistance of AI， medical examination and diagnosis will be more effi cient and accurate， which will allow patients to acquire high-quality medical service in a more convenient manner，” He Chuan， CEO of Proxima， said.
According to a recent report by the journal Nature Medicine， researchers have developed a new machine-learning program that can not only confirm the type of lung cancer with 97 percent accuracy， but also detect the mutant genes that cause cells to grow abnormally.
Apart from imaging diagnosis， AI technologies such as natural language processing， big data analysis and robotics also aid other applications. These range from maintaining electronic medical records to surgical robots， wearable devices and drug research， innovations which could transform the whole medical industry in the future.
The use of AI has made the healthcare industry the next gold mine for investors. According to CB Insights， a market research organization， healthcare has become a key area for the AI industry in terms of research and application. Global startup companies in this fi eld have attracted nearly $4.3 billion financing since 2013， surpassing the other industries supported by AI.
China’s use of AI in healthcare， though a late starter， has “exploded” in recent years， Eliot Siegel， Chairman of the Medical Resource Image Center at the Radiological Society of North America， told Chinese newspaper 21st Century Business Herald. “Its clinical application in certain areas has outperformed the United States’. Additionally， China’s overall performance in the AI industry shows little difference from the United States’ and could be the most likely to transcend America’s，” Siegel said.
Adding to resources
AI is just what the doctor ordered for the world’s most populated country. According to China’s National Health Commission， of the around 1.4-billion population， only 12 million are medical workers. Less than 10 percent of the hospitals in China have advanced medical facilities， according to a report by the State Information Center. Moreover， there is a signifi -cant gap between large cities and smaller ones， as well as urban and rural areas in the distribution of high-quality medical resources.
In the face of insuffi cient medical resources and the unbalanced distribution， AI can improve the operational efficiency of top hospitals， maximizing their service capacity. Deep learning or machine learning can use machines and software to duplicate some of the skills of doctors from top hospitals and apply them to help patients in areas with less developed medical resources.
Realizing the significance and necessity of developing intelligentized healthcare， the government has increased its policy support for the sector from 2016. In July 2017， the State Council， China’s cabinet， released a guideline to develop AI technologies， making its use in healthcare a pivotal task. AI was also taken into consideration in the central government work reports in 2017 and this year， making it part of the national strategy.
“These policies are the driving force for developing the AI industry，” Shang Yang， a medical industry analyst with Yiou， an AI think tank， told news weekly Oriental Outlook.
With policy patronage and growing market demand， capital has been pouring in. As of August 2017， the cumulative financing received by China’s AI healthcare companies had hit 18 billion yuan （$2.62 billion）， according to Yiou. A report by Chinahyyj.com， an industrial research website， put the market size of China’s healthcare AI at 9.66 billion yuan（$1.41 billion） in 2016， marking a 37.9-percent increase year on year. This year， the amount is expected to top 20 billion yuan （$2.91 billion）. Baidu， Alibaba and Tencent， three of China’s largest tech companies， have seized the opportunity. Tencent Miying， launched in August 2017， focuses on medical imaging and auxiliary diagnosis. It has gained substantial traction for early screening for cancers and is working with nearly 100 hospitals as a leading AI innovation platform in the healthcare industry.
So far， so good. But the use of AI technologies in healthcare has its challenges too， especially for large-scale application.
The data challenge
To apply AI in real life， it’s vital to have suffi cient data， Yiou’s Shang told Guangming Daily. Though China’s overall volume of data is large， its quality is not high enough. Take medical imaging， for example. The data collected must be corroborated by doctors with extensive experience before being fed to machines for machine-learning. However such resources are far from enough.
Data security is another major issue. In China， medical data belongs to patients and hospitals. So AI companies have difficulty in acquiring data. Also， a patient， especially when diagnosed with a complex disease， may go to several hospitals. If AI companies obtain data from only a single source， its value would be marginal.
Wang Zhenchang， Vice President of Beijing Friendship Hospital affiliated to the Capital University of Medical Sciences， told Oriental Outlook that currently most AI companies are establishing servers or clouds in the imaging departments of hospitals. He said the action poses a huge potential safety hazard and could undermine the national requirements for network security.
To cope with this， the National Health and Family Planning Commission， predecessor of the National Health Commission， issued a plan in 2017 to make medical data open and shared. The policy has led to some companies establishing medical databases.
On September 13， the National Health Commission also unveiled new policies， encouraging medical institutions and corporates to share medical data while also emphasizing the need for them to address risks together.
Besides data sharing， the involvement of doctors can help AI companies better understand the workfl ow of hospitals so as to create products that suit real scenarios. This would then lay a solid foundation for the large-scale application of AI.
Wang has more requirements. “Clinically applicable products must be simple and user-friendly for doctors，” he said. “This is essential.”
“Healthcare is highly specialized， so the application of AI in healthcare can’t survive without the deep engagement of the medical community，” Shang said， noting that interdisciplinary talents well-versed in both medical sciences and AI technologies are “the scarcest resources.”
However， the growing use of AI is not going to lead to the apocalyptic science fiction scenario where machines rule man. At least not in the immediate future. As Zhou Xiang， co-CEO of United Imaging， a medical device company， put it， “In the foreseeable future， AI would not replace doctors， but doctors who know AI technologies would replace those who don’t.”
The booth put up by Proxima， a startup providing medical imaging by using artifi cial intelligence （AI）， demonstrated hi-tech virtual diagnosis where visitors lined up for a roleplaying game， being a doctor of the future. By simply clicking on an operation panel， these “doctors” could watch on a big screen how AI technologies facilitated medical examinations， radiography and diagnosis， even coming up with therapeutic suggestions， and all of this within seconds.
The lab displayed the virtual intelligent diagnosing of complicated medical problems like lung cancer. Once a “doctor” chose the lung module on the panel， a three-dimensional scan of the patient’s lungs appeared on screen. A red circle immediately marked out the nodules where the cancerous cells were， a hi-tech method that can save doctors valuable time， making an early diagnosis and pinpointing the areas where the disease has spread.
“With the assistance of AI， medical examination and diagnosis will be more effi cient and accurate， which will allow patients to acquire high-quality medical service in a more convenient manner，” He Chuan， CEO of Proxima， said.
According to a recent report by the journal Nature Medicine， researchers have developed a new machine-learning program that can not only confirm the type of lung cancer with 97 percent accuracy， but also detect the mutant genes that cause cells to grow abnormally.
Apart from imaging diagnosis， AI technologies such as natural language processing， big data analysis and robotics also aid other applications. These range from maintaining electronic medical records to surgical robots， wearable devices and drug research， innovations which could transform the whole medical industry in the future.
The use of AI has made the healthcare industry the next gold mine for investors. According to CB Insights， a market research organization， healthcare has become a key area for the AI industry in terms of research and application. Global startup companies in this fi eld have attracted nearly $4.3 billion financing since 2013， surpassing the other industries supported by AI.
China’s use of AI in healthcare， though a late starter， has “exploded” in recent years， Eliot Siegel， Chairman of the Medical Resource Image Center at the Radiological Society of North America， told Chinese newspaper 21st Century Business Herald. “Its clinical application in certain areas has outperformed the United States’. Additionally， China’s overall performance in the AI industry shows little difference from the United States’ and could be the most likely to transcend America’s，” Siegel said.
Adding to resources
AI is just what the doctor ordered for the world’s most populated country. According to China’s National Health Commission， of the around 1.4-billion population， only 12 million are medical workers. Less than 10 percent of the hospitals in China have advanced medical facilities， according to a report by the State Information Center. Moreover， there is a signifi -cant gap between large cities and smaller ones， as well as urban and rural areas in the distribution of high-quality medical resources.
In the face of insuffi cient medical resources and the unbalanced distribution， AI can improve the operational efficiency of top hospitals， maximizing their service capacity. Deep learning or machine learning can use machines and software to duplicate some of the skills of doctors from top hospitals and apply them to help patients in areas with less developed medical resources.
Realizing the significance and necessity of developing intelligentized healthcare， the government has increased its policy support for the sector from 2016. In July 2017， the State Council， China’s cabinet， released a guideline to develop AI technologies， making its use in healthcare a pivotal task. AI was also taken into consideration in the central government work reports in 2017 and this year， making it part of the national strategy.
“These policies are the driving force for developing the AI industry，” Shang Yang， a medical industry analyst with Yiou， an AI think tank， told news weekly Oriental Outlook.
With policy patronage and growing market demand， capital has been pouring in. As of August 2017， the cumulative financing received by China’s AI healthcare companies had hit 18 billion yuan （$2.62 billion）， according to Yiou. A report by Chinahyyj.com， an industrial research website， put the market size of China’s healthcare AI at 9.66 billion yuan（$1.41 billion） in 2016， marking a 37.9-percent increase year on year. This year， the amount is expected to top 20 billion yuan （$2.91 billion）. Baidu， Alibaba and Tencent， three of China’s largest tech companies， have seized the opportunity. Tencent Miying， launched in August 2017， focuses on medical imaging and auxiliary diagnosis. It has gained substantial traction for early screening for cancers and is working with nearly 100 hospitals as a leading AI innovation platform in the healthcare industry.
So far， so good. But the use of AI technologies in healthcare has its challenges too， especially for large-scale application.
The data challenge
To apply AI in real life， it’s vital to have suffi cient data， Yiou’s Shang told Guangming Daily. Though China’s overall volume of data is large， its quality is not high enough. Take medical imaging， for example. The data collected must be corroborated by doctors with extensive experience before being fed to machines for machine-learning. However such resources are far from enough.
Data security is another major issue. In China， medical data belongs to patients and hospitals. So AI companies have difficulty in acquiring data. Also， a patient， especially when diagnosed with a complex disease， may go to several hospitals. If AI companies obtain data from only a single source， its value would be marginal.
Wang Zhenchang， Vice President of Beijing Friendship Hospital affiliated to the Capital University of Medical Sciences， told Oriental Outlook that currently most AI companies are establishing servers or clouds in the imaging departments of hospitals. He said the action poses a huge potential safety hazard and could undermine the national requirements for network security.
To cope with this， the National Health and Family Planning Commission， predecessor of the National Health Commission， issued a plan in 2017 to make medical data open and shared. The policy has led to some companies establishing medical databases.
On September 13， the National Health Commission also unveiled new policies， encouraging medical institutions and corporates to share medical data while also emphasizing the need for them to address risks together.
Besides data sharing， the involvement of doctors can help AI companies better understand the workfl ow of hospitals so as to create products that suit real scenarios. This would then lay a solid foundation for the large-scale application of AI.
Wang has more requirements. “Clinically applicable products must be simple and user-friendly for doctors，” he said. “This is essential.”
“Healthcare is highly specialized， so the application of AI in healthcare can’t survive without the deep engagement of the medical community，” Shang said， noting that interdisciplinary talents well-versed in both medical sciences and AI technologies are “the scarcest resources.”
However， the growing use of AI is not going to lead to the apocalyptic science fiction scenario where machines rule man. At least not in the immediate future. As Zhou Xiang， co-CEO of United Imaging， a medical device company， put it， “In the foreseeable future， AI would not replace doctors， but doctors who know AI technologies would replace those who don’t.”