AI for Good Webinar #3: China’s digital health technology in combating COVID-19
7:42AM Apr 15, 2020
Uh Hi Good, good morning, good afternoon, good evening. This is Shan from CAICT and it is really a pleasure for me to have the opportunity to share something that we are using the digital health technologies in combating Coronavirus in China. So this is my slides and uh.
Yes. And uh, ah, so, so, as this webinar is open to all and no very specific target audience, so I prepared a very vivid way to share our experience with some digital health use cases. In uh my contents consists of this three part first, a brief introduction of the background, and then some typical ICT cases in this outbreak and finally some further discussion and conclusion.
So this is the start of the outbreak. You probably know that and uh is that it is about on the December 13th, an unknown virus appeared in Wuhan and this is, this picture is our national TV report on the next day, experts from National Health Committee arrived in Wuhan at a morning. So this is Thursday. Daily change curve and zeros quick response from Chinese government. And I attached a link of WHO and China joint mission press conference on this page. So if you are interested in on the details of this, of this picture, you can find more information from that. And during my slides, I just want to say that as the as the graph shows, you may expect curve continue climb and as the red line, but we see peaks here actually. So you may ask on how about the situation is going on, because this graph is only updated on February 20th. So, just as it's shown as a blue line explained as of April seventh. According to the statistics of NHC, the daily confirmed cases are 62. And most of them are international friends. So we can say that things are quite on the control now, and uh, and and also we can see that the difference between the red line between the red line and the blue envelope, we can measure that part as the effect of the prevention and control measures. So, what does this difference and also the effect come comes from and or what is the percentage of ICT in it? The answer from my understanding, I think digital health plays a very important supporting role but not the main part. In order to be honest, we have to give credit where credit is due and we really appreciate the hard work from the medical staff at a frontline, you know, during this outbreak over 14,000 Medical workers from all over the country, risking their lives and are also saving others life as the front line. And also the infrastructure workers. They work, they're working day and night during the New Year festival, making a ten-day hospital construction possible. And also the government's quick response from the very beginning and to the home process, and are also the manufacturers supporting the production, production, medical production, and as well as the community care, community workers, and also volunteers, and they're timely but important actions just to sum up to make a huge change. To protect the lives around us. So we can say that it is the result of everyone working together.
So, as a result, I really liked there's a statement from who we must remember that they are not numbers, they are not numbers, and not not just at the point on the curve, they are people living people with their same motivations and feelings like every one of us. So if we see things from that perspective, we will understand the unbelievable hard work and urgency to make to take actions as their home community. And as an ICT engineers, we would probably also to have our contributions and we may start with this question. What Kind of contributions that digital health could do. And I think one common answer could be that to make life better. So here the the word life in that answer I think could be everyone that we have mentioned in our cases that worked on the patients, doctors, manufacturer, and construction workers and their citizens and their volunteers and so on. So as their growing part shows, with various available technologies, we can actually play a very positive supporting role in fighting with their corona virus. So here comes the I call it magic combination and we just group um... regroup what we have the demand and also the gray part of the digital health part. we regroup them as a puzzle. And I think it is a very important way to start with this, just to stick on the application scenarios, and to feel the demand. And before, we're just finding a fancy solution and products. So, we would like to use technologies like the magic tool, and we want it to go to where it is in need. So, just let me take an example. And from what we what I have mentioned before, we understand their heavy load of the frontline doctors and we want to help to lace the burden on patients treatment. So there could be an assisted diagnose system with AI Also the telematics system based on 5g and if we want to support the production of the phone line worker, robots can also be a very, very good tool to do deliver, bring and also disinfection work in Meanwhile, the effective production and also dispatching in could also be a strong support and of the of the frontline workers. And besides their case tracking in their statistic, platform and also public and public place sharing could also actively help with the risk assessment and also the risk communication and to protect the ordinary people then to reduce their home risk level of their home society. And last but not the least, with various applications to meet basic living this could also support the lockdown life and to make the quarantine life be more easy. So this feature acts as an index of the following use cases that I'm going to share and I will give you more specific description later.
So, according to what I have mentioned, the telco their digital technologies are just regrouped in orders. And the first case that I want to introduce is telemedicine that based on 5G, and er This is a picture that on February 9, the PLA General Hospital conduct a remote consultation based on connected to the first 5g network with Wuhan emergency hospital. And we can see that experts are there and electronic screens in front of them, and the results of patience and of the medical records and also the indicators are uploaded in real time and with the highest resolution image and we know that they are very clear at a glance. And actually this kind, it is a kind of practice that already tasked and used in China before the outbreak. And if you are interested, you can also search for more news report that before the outbreak that we have the remote surgery last year. And so that is why 5G and telemedicine technologies response to the corona virus is so fast in China. Because it has already been tested and used. And because of the national mode and infrastructure possibilities. So we can see also see it as their kind of preparation on the digital health part before the outbreak. And this time they also made a little innovation on the terminal. So we call it 5G telemedicine cart. This terminal is removable and could contain some medical supplies below the video. So in the use cases with 5G experts from both sides, we can see that in the picture from both sides could shared the patient's medical files with high-definition image and it could definitely enlarge supply of the front line expertise and also effectively reduce the infection on the medical workers. And besides 5G and telemedicine, AI is also a kind of strong tools could help in this prevention and control of the outbreak. Just take one product as an example, the infrared, and they're also actually topic driver in our AI for health Focus Group. And their system takes just a few seconds to process dozens of CT scans and provide support for rapid screening and also give some suggestions to doctors. And this is a general workflow that I always used in our focus group to say something about the AI for health process. And we can say it from a different perspective. From at the bottom from the users perspective, we can see AI as a kind of black box in that is the structure or non structured data put inside and done with their strong computing
possibilities. And if we test with an answer, many based on correlation, so it could, the answer could be some screening without the diagnosis, reference and also and health management suggestions and so on. And also to make their blackbox be reliable and safe, we also have another kind of perspective. In the industry, we also have to do it within regulation, right or regulation to commit the regulation requirement, for example, the benchmark assessment and also the clinic trials. And also a very important part is that the post market surveillance to help the overall AI system be safe and trustworthy. So what I have mentioned to diagnose with the corona virus is just one kind of example of AI for health applications. So if you are interested, you're welcome to click our Focus Group links which I have attached here and there you can find more details. And also, um, apart from the diagnosis in the clinical line, AI is also with great potential to support the vaccine development. And you might probably heard about the double twenty principles in pharmaceutical industry. It is a kind of saying that you should have an average time of 20 years and also an average cost of 2 billion US dollars to develop a new kind of drugs on the market. So this kind of two numbers are in the industry to emphasize how difficult it is to develop a new drug.
And this picture, what I have, I want to show that the AI also help with the vaccine development and this picture is a kind of modeling of some target for the proteins sequence of the corona virus. It is based on the Alibaba cloud, to develop a drug RND platform and to support the vaccine development for corona virus. And also, Ma Yun, the founder of Alibaba also launched a donation of 100 million and opened the cloud computing capacities to support the development of vaccine for corona virus. The details are on the slides but what I want to mention is that you have to understand AI just help with several steps of the upset development work and just for the compound testing and screening. So after they're discovering candidate drugs, you also have to go through what we have mentioned before their preclinic research trials and also to apply for a new drugs and then get approval from the regulations. Then you can go to the market. So, according to the latest news, China and the EU as a first step into the clinical trial paces in there, so we are also very looking forward, what is going on next and hopefully, the vaccine could be developed very soon. So additionally, AI could also help with the
public place screening.
Especially during the societal recovery stage in the public place screening could be very helpful and important. This picture just show a kind of long connect a temperature detection system, it could be used in very densely populated areas such as the railway station, the airport and the building entrance and etc. And then the next page shows some process and the implement and I know Ms Yuan will have more technical details in her presentation. So I'll just skip it and leap in. There are more details in this presentation and your I know that The last time and our European colleagues also share some tracking in case tracking in use cases on on the corona virus. And the China's CAICT, China Telecom, China Unicom and also China Mobile on jointed and launched travel card based on their telecommunication data and to just give you a self check and also the proof if you have ever been to an epidemic region in the past 14 days or not based on the the base station data so it's a kind of a traffic light. For example, if you are no risk then we give you a green card. And the middle risk, you have a yellow card and high risk for a red card. So the analysis on the back end is based on the majors three telecommunication operators. It is a kind of service that provide the individual risk assessment for 1.6 billion mobile phone users across the country. And there it is a kind of one-click inquiry of the countries and also regions that you have visited within the 14 days to give you proof and also their risk assessment result. And for the fifth case that I want to share is about the Medical material dispatching. So, um, so this picture is that on February 1, Premier Li came to check the national key medical material dispatching platform on the public on the production and also the dispatching of their protective stuffs. And we also have a platform that developed it by MIIT the Ministry of Industry and Information Technology. And on that platform we have 21 categories of their key medical supplies, and most importantly, we have the function of the monitoring on the production capacity and also the scheduling and also the output of all kinds of the medical supplies. And also
besides that part of the platform, CAICT also developed a posting platform at the very beginning of the outbreak. And with that platform that I have showed in the slides, you can post the demand information and also the supply information as well as your contact to just talk the supply and demand of the digital health resources. And this platform is also approved by the joint defense and control mechanism of the State Council. So the products that you could post includes the medicals protections software's There are solutions on intelligent devices equipment and also what we have mentioned like Alibaba as a computing capacities and so on. So this is the virtualization of this kind of digital health resources supply and or demand platform and also the risk on communication as another very important use case that I want to share. So no one is safe until everyone is safe. So during this outbreak just to be transparent in time and accurate to let everyone receive the right knowledge to protect themselves is very important. So take the National Health Committee website for example, it act as their authority channel to daily update The statistic data and also promote the protection, knowledge and clinical strategies. And also we also have homepage that embedded in recharge and developed by the yuen in their, you know during their lockdown day, my daily routine is to check for the real time updates on that platform and we can see the curves and also the maps and also related technology also related knowledge on that page. In the as far as I know, they also developed an English version to support the global rescue. And there for the seventh use case that I want to share is about the multifunction robots. So this is a picture of a medical robot that was officially used in Wuhan emergency hospital. So with the function of disinfection and also delivery. It can reduce the workload of medical stuff and also the risk of cross infection in hospital. So actually, there are also other kinds of examples of the robots, for example, the temperature measuring, disinfection, and also the delivery robots in different shapes and developed by different countries. We can also see the applications of robots in our daily life before the outbreak, but during this very special period, the role of the robots has been greatly amplified and also recognized by public. And finally this is the last use case that I want to mention. It's about the various living applications that to support the lockdown on life for most of the people. For example
the online consultations, online shopping, online education & works and all sorts of psychological interventions and so on. So, this is just a few examples that I have it in this page, for example, the community entering and exit restriction system and also the online consultations. And also the you order your take-away online including groceries and also some teleconference and also the online educations to make you able to work from home and study from home and also some other entertainments that based on the digital technologies.
So, that is all use cases digital use cases that I want to share in my speech and if you are interested in more, we actually CAICT actually published a kind of best practice on the digital health cases in corona virus. And for now we have over 117 cases in our reports and there will be continuous updating in our website and there's English version of the best practice report which is on the way. So, for the last part that I want to say is that we are not limited and just willing to do unlimited part of research on just the use case is the perspective or use cases, we also want to monitor and evaluate the effects and of course different digital technologies. And so, in that way we believe we can have more valuable experience that we could share to different countries. So then they can, based on our experience, pick what is the suitable way, the strategy for them to do to fight with the corona virus. So for this part, we as a start, we designed a questionnaire on the monitoring and evaluation part. So we will very much appreciate if you can share some information from your side and to give us some input in this questionnaire, and also you're welcome to contact me if you are interested in the evaluation part. So this is the summary page. First, I want to say that digital health will not be the main contributors in combating Coronavirus, but it could play a very important supporting role on the control and prevention work. And for the second part, I share some digital health use cases, for example, on the telemedicine AI for health in the platform on the medical material dispatching and also the risk communications and also robots as well as some digital living applications to support most of people's life. So more details you could find In our CAICT best practice report and for the last part, we want to share reproducible experience for other countries and we realize the importance on the monitoring and evaluation of the effectiveness of different digital health cases. So, we will be very appreciate if you have some input and also some comments on that part also as well as the questionnaires. So, as the last I want to thank Ms Xu Weiling, Ms Liu Rui, Ms Zhang Xueli, Ms Min Dong, Ms Wang Yapeng for reading the first version of my slides and giving important guidance. As well, I want to thank for the cases and data supporting from Ms Liu Yanfei and other colleagues from the CAICT as well as continuous supporting from Dr. Luo Zhong the Chair of SG16 of ITU. And also thanks to Ms Gauden Galea and Ms Mengji Chen from WHO China office to provide very valuable suggestions on our research. Most importantly, thanks to ITU for giving me the opportunity to share this information and also to further discuss with you. So thank you for listening. That is all of my slides. Thank you.
Good morning. Good afternoon and Good evening, everyone. This is Yuan Zhang from China Telecom, Machine Vision Standardization and Strategy Department. I'm the rapporteur of Question 12 in Study Group 16, and also the co chair of MPEG video coding for machines Ad Hoc Group. And we're honored to have this opportunity to share some of our working on COVID-19 combatting. Let me share my slide. And Shan has just made an impressive presentation of the pandemic combatting. We have some similar work and also joint work with the CAICT as well. I'll continue to introduce our work in the perspective of the telecommunication operator. So my talk includes three parts. First, our role in pandemic combatting and then observations and facts and then the measures that we have taken. Shan has just gone through background and timeline about the outbreak in China. I'll skip this part. So now we are facing a global crisis and unexpected test for humankind. It's a test for our health care system, our society, our humanity, and of course, our telecommunication systems, as people are locked down or quarantined, and we are trying different approaches to pass this exam to survive. The decisions that we made during this period of time, we all contribute to the reshaping of the world for years to come. Things are different among countries, although we're in different battles, but we are in the same war. It's against the time and the virus.
So we just try to take the most effective methods in most efficient way trying to balance the trade off. So I hope our experience will be of use as China is launching this first. What role do we play in this pandemic where no doctors or nurses, no policymakers so we're notching the frontline. As a telecommunication operator, our advantages first of all rely on our telecommunication infrastructure. We have more than 1 million 4G base stations, and 75,000 5G base stations, and about 170 million broadband subscribers. And also we have a close link to our customers. We're still providing on site and on spot service during the pandemic. And our goal is to transform technology into capabilities for continuous work study in life and also the epidemic prevention during this pandemic to 10s of thousands of households, as the big company we will shoulder the social responsibility by providing infrastructures and services, both for health care and for quality of life. So based on this, our role and our mission, we take measures accordingly. Next, I'll share some of our observations and facts. I looked into our status check data in February, the requirements for remote application are increasing. Regarding business to business services, the most visible ones are remote we do applications. Among all of the emergency applications. The top two are health care and education associated applications. For example, the hospital Information system, online schooling, and we have launched the cloud conference in January within one month. After launching, we have got more than 700,000 subscribers which is 22 times growth compared to data before pandemic. And regarding 5g applications, we have supported the live crowd overseers which, which is an interesting project. I will talk about that later. And the telemedicine as Shan has just introduced, and the top business to customer services, online education and telecommuting, though this is our first observation, which seems obvious. And let's look at the second one, the network flow is increasing. We can see from the statistics here that the wireless and broadband network flow are both increasing about 20%. But the QS didn't drop.
Then you might want to ask what approaches do I take. The data transfer speeds and download speeds and now slowed down and everyone can still access to the internet well. Smartphone just as well even in Wuhan city. So what did we do? Stay at home policy will definitely add to network congestion. I've learned about that mobile operators from some countries have asked the user to reduce their data consumption. And also there are some novel ideas such as providing 4G signal from a high altitude balloon, which is very interesting, and also asking the bit streaming services such as Amazon, YouTube, to reduce their quality of videos to free up their capacity. And also, some countries granted networks additional radio spectrum on a temporary basis. But we didn't take those approaches. I tried to give some explanation here through comparison. We don't have significant drop of QoS for 4G LTE, IPTV short messages, etc. The reason is simple. We could take a look at the figure in the table. The total of broadband subscribers in China until last August is more than 400 million and fibre to home subscribers also exceeds 400 million and broadband subscribers with a bandwidth more than 100 megabits per second is about 350 million. So, what does this say? It means that the peak level of data consumption did not come close to the limit of the broadband network. So we have advantages in optical access networks and fibre to home deployment. Persistant efforts to put network has been paid off. China Telecom is continuously enhancing our broadband access bandwidth 100 Mbps is the baseline bandwidth now. And the bandwidth more than 100 to the home accounts for about 80%. I'm not saying that why you took no measures while you have constructed base stations and also optical networks in certain regions, especially in Wuhan to increase capacity. And there's one more thing to be noted that QoS does not only depend on the access network bandwidth, but also the bandwidth of backbone. So the QoS at night is not as good as daytime because the traffic during night time is easily congested especially for the international gateway.
And there are some other facts and observations based on statistics on fiber array. Data roaming users and data volume are dropping due to the travel restriction and risks. The volume of short messages is decreasing which has been a trend as IM applications are booming except for public services, short messages, a lot of short messages are sent about COVID-19 buy different forms or applications and there are increasing need for VPN remote access as a result, city requirements are surging. Although we didn't stop on site service, there is a need for service online and there are growing need for CDNs (Content Delivery Network) with requirements of 95 peak value of 3 terabytes and total flow of 45 petabytes. Virtual cloud server requirements and use ratio of the resources has increased. And the cloud computer is developing and wanting to use. Blowout out of remote applications with novel requirements, most of them are not a new invention such as remote conferencing, but during work at home, we found that the function of beautifying and clothes changing would be very helpful. As maybe we would like to have the function of clothes changing while we work at home and maybe in bed. And there are a lot of products and solutions which are the integration of capabilities to better serve the requirements and effecttively China Telecom has developed more than 17 new applications for different regions and industries. And now I'll introduce some of them in the next part. So the third part measures that we have taken. The miracles just Shan mentioned that we have build a hospital in 10 days, there is something more I could talk about. Instead of one, we built three. The Huoshenshan hospital, Leishenshan hospital, and a special hospital for women and children, all in Wuhan city. And we were building them on cloud. I mean, all the medical systems and the information systems are deployed on cloud. They are the first hospitals on cloud in China. The whole construction took 10 days, but we only took three days to deploy the network and cloud the resources. So what's on cloud, it's normal to have the accelerate information systems on cloud, but the Huoshenshan and Leishenshan hospital applying the our on-cloud solution, so major business and information systems are the Hospital Information System (HIS), Laboratory Information System (LIS) and Picture Archiving and Communication Systems (PACS). They're all on cloud and so other operation management, resource management, knowledge management, customer service, including the queuing system and all databases. The measures we took that for one hospital, we have seven dedicated cloud server to support the service and applications and we're deployed the same type of services in different underlying servers, physical servers. We designed a host and backup database and backup the data timely to make sure the availability and reliability. The benefit of this hospital on cloud is that IT maintenance in local is reduced. We only need to talk and test in the given environment. And also the deployment of our healthcare applications accelerated. And we could think about duplicating the solution. Actually, we did the Huoshenshan hospital first and then we copied their solution to Leishenshan hospital. And we could expect that it could be shared among hospitals in future. It's a potential standardization area with common interfaces and data formats, systems and data could be shared under certain rules in future.
And the interesting thing about this, we have launched 24 hour high-definition live broadcast of the construction of the two hospitals. The highest peak of online users exceeded 100 million. It is called in China the strongest cloud formation cloud overseers project in history. And it's based on 5g network and our cloud platform. While people are quarantined at home and cheering up for workers on site remotely, we might have a new record here. Another major application is the track querying app, which provides the individual subscribers with functions including regional risk query, epidemic situation forecost query return to city report query, itinerary and content query. And on the basis of ensuring user privacy and safety, it conducts real time perception of the flow of people in key epidemic areas, with cooperation with relevant provinces. And it can also provide interfaces to other stakeholders while cooperating. It's about supporting CAICT on the case tracking and modeling app it get people's location and analyze it that people flow in key academic area, such as favor clinics and people gathering areas to provide open and transparent data and dispel the unnecessary fear of the epidemics. About 20 billion signal data and billing data were processed every day, covering our 2g, 4g, 3g users. And to the end of last month, there are 70 million query records. The track query analyzes information such as cities and countries that the user has passed or stayed within 14 days through historical data. So people can query potential risk and know better of their own status. It has introduced AI capabilities to develop prediction models analyze the possible future development trends of the academic to realize dynamic factor adjustment and the ability of self predicting the epidemic. Individual users can inquire about their own itinerary, the possibility of corona virus user contact and regional risks through their mobile phone number and understand the risk in real time. The products have been currently used by more than 10 million people. There are promoted by China Telecom as a key application service product serving the public in relevant provinces and cities. This app also have a function of epidemic prediction. The main data that the infectious disease dynamic model relies on all from authorities, which objectively describes the changing of the academic situation and provide a basis for epidemic prediction. The application has introduced an important factor population mobility. As population mobility is the main factor in the development of the infectious disease the OIDD data is used to calculate the frequency of population mobility and to optimize the prediction accuracy of the model. So the common models include SI, SIR, SIRS, and SEIR models. The SEIR model is mainly used to predict the memory of infectious diseases with latent period, which is consistent with the characteristics of COVID-19. We have achieved the accuracy of more than 99% for existing confirmed case and also more than 99% for the total confirmed case. So, just now Shan has mentioned about a similar app regarding the temperature detection. We integrated this function to our surveillance system through black body and infrared gray in the front end devices to precisely detect the temperature and real time alert software in the terminal devices to enable the real time temperature tracking in areas with heavy traffic from a passenger flow like train station, metro station and entrance to shopping mall etc. And mask recognition algorithm is implemented in certain scenario like in closed area that people are asked to wear masks to accessing in. It will also send an alert when the camera detects people who are not wearing masks appropriately and also when the density of people exceed the maximum value.
And as well we have a lot of other applications which I will not go into detail. Some of them are interesting. Like we have the unmanned business office for people to subscribe to our services and the 5g you have a unmanned aerial vehicle with thermometers and intelligent loudspeakers to patrol and find people with fever and to remind...
Alright, disconnected for a minute for a second. I will continue. And to remind people of no gathering and we have a lot of other emerging applications. We have the unmanned business office the 5G UAV and if you are interested, you could contact me for details.
So, this is all my presentation. Thank you for listening. And I would like to thank all my colleagues for help to gather the materials. Please feel free to contact me if you have any question or something you would like to have further discussion. I'll try to answer them or forward them to our experts in particular areas. And you can also contact me if you are interested in machine vision associated standardization work. Thank you very much.