Using AI to control COVID-19
There is no greater challenge to the public health system than a pandemic, such as the novel coronavirus outbreak that emerged in Wuhan, China.
For a fast-spreading virus like COVID-19, time is of the essence and China’s experience in quickly converting AI technology into practical applications has proved useful. AI-powered applications are being used to combat the virus by speeding up screening, diagnosis and new drug development.
AI is used in infrared cameras to scan and screen people en masse for high temperature at airports and railway stations across China. Together with facial recognition cameras, the cameras can identify the person with a high temperature, even when wearing a surgical mask.
AI was used soon after the outbreak was discovered to help sequence the new coronavirus genome, so that research on a vaccine could get started. In the previous SARS outbreak in late 2002, scientists had to wait months before they knew what the germ looked like. With COVID-19, it only took Chinese scientists using AI a few weeks to sequence the virus’s genome before sharing it online with researchers around the world.
China’s experience in quickly converting AI technology into practical applications has proved useful in tackling COVID-19
Knowing the genome sequence allows researchers to design lab tests to identify the presence of the virus. Unfortunately, the new COVID-19, like SARS and MERS, is also a single-stranded RNA virus, which means it is susceptible to mutation and harder to test. Using AI, Alibaba was able to counter this problem and reduce lab test time to diagnosis coronavirus infection from hours to only 30 minutes, with a new ‘whole genome detection’ approach.
Besides lab tests, a CT scan of the lungs is also an effective way of detecting signs of coronavirus infection. During an epidemic, radiologists are overwhelmed with thousands of scans to inspect each day. China turned to AI to automate this process. Alibaba trained an AI deep learning system using thousands of CT scans from confirmed cases. The resulting AI model was then able to analyse a CT scan within 20 seconds with 96 per cent accuracy of detecting COVID-19.
AI companies in China, such as Alibaba and Baidu, have been offering free and open source AI technology, datasets and AI compute power to public research institutions around the world to help shorten R&D time in combating the disease. Baidu also open-sourced its AI LinearFold algorithm that predicts RNA virus 3D structures and behaviours and allows scientists to understand better how the virus invades our cells so that matching vaccines can be created. Using LinearFold, prediction time can be reduced from 55 minutes to 27 seconds. Alibaba also made its AI drug discovery platform available to researchers. Historical data on drug R&D efforts for coronaviruses, such as SARS and MERS, allows AI prediction models to provide insights on how existing drugs might be repurposed for COVID-19 and thus improve the efficiency of new drug screening.
AI chatbot technology is used to help reduce pressure on hospital and government frontline personnel by automatically answering inquiries from the public. AI chatbots also provide advice to individuals on whether they need to go to a hospital for screening or stay at home for the 14-day quarantine. In Shanghai and other cities, AI chatbots automatically call high-risk patients to assess their condition. Physical AI robots are also used in segregated wards in Shanghai hospitals to disinfect those wards, as well as deliver food and medications and check body temperatures.
Facial recognition is used to ensure affected citizens are following the 14-day self-quarantine requirement. According to reports, people who violated the quarantine and did not stay at home were automatically detected by facial recognition cameras and contacted by the police as well as employers. A Chinese company created an AI system that allows users to check if they have recently travelled with someone who has contracted the new coronavirus. The system uses public data and other information sources to correlate with an individual’s travel itinerary. It was reported that more than 21 million people used the service within two days of its launch.