The COVID-19 Pandemic And Taiwan’s AI Development

Taiwan AI Labs has developed an AI tool that screens chest X-rays and determines if a patient has pneumonia or is infected with SARS-Cov-2. Photo: Taiwan AI Labs

The COVID-19 pandemic is taking an unprecedented toll on the global economy and human lives. Yet it is also inspiring the innovative application of new technologies to research the disease’s origins, treat it, and manage its spread. Nowhere has this been seen done more effectively than in Taiwan, where the government has used AI and Big Data analytics to track people under enforced quarantine and trace the movements and contacts of infected individuals.

Research institutions, academia, and industry in Taiwan are also getting on board, developing AI algorithms to target specific issues that have arisen as a result of the pandemic. The products and solutions that have resulted from these efforts have the potential to benefit countries the world over.

Taiwan AI Labs, a private research organization that works closely with academia, government, and industry, is currently working on a range of AI-based solutions to combat COVID-19. All of these solutions are open source and available for public use at the web address covirus.cc. They include a platform that uses genetic sequencing data from the German GISAID database to map all of the different SARS-CoV-2 strains with phylogenic trees. The resulting visual aid shows the full evolution of the virus and can help doctors and researchers trace a patient’s viral strand to its origin.

“Using this tool allows us to understand the relationship between virus strains,” says AI Labs founder Ethan Tu. “And we can analyze in detail how this virus has mutated. This helps us detect the source of the virus in each patient – whether it is from the U.S., Europe, or elsewhere.”

The covirus.cc website also hosts a medical image screening platform that uses deep learning to detect coronavirus in a patient’s chest X-rays. The algorithm has been fed over 40,000 chest radio-  graphy images, including around 5,000 images of patients infected with SARS-CoV-2.

Tu says that AI Labs is now working with the National Health Insurance Administration to implement the platform in Taiwan. They have installed the algorithm in the NHI’s central database to automatically screen all chest X-ray data uploaded from hospitals. The software alerts doctors if the X-ray is found to present pneumonia or SARS-CoV-2. Having this system in place, says Tu, will eventually help shorten the waiting time for getting a diagnostic result from one week to a matter of minutes.

Another project AI Labs has launched on covirus.cc is a drug screening model that tests the efficacy of possible COVID-19 treatments, including the anti-malarial drug hydroxychloroquine touted by U.S. President Donald Trump. The accuracy of the technology allows biomedical labs to determine the priority of drugs to bring to clinical trials.

Tu says the model, which only tests drugs already approved by the Food and Drug Administrations in Taiwan and the U.S., produced screening results on a range of treatments as early as February 4. Included among the drugs tested was the antiviral Remdesivir, which the model found would be useful in treating COVID-19.

Among the AI-related COVID-19 research projects the semi-governmental Industrial Technology Research Institute (ITRI) is working on is a prediction model to test different approaches to “flattening the curve” – reducing the amount of infections in a population. The model, which ITRI is developing in collaboration with the National Applied Research Laboratories, will analyze methods such as mask-wearing, social distancing, quarantining, and others to determine which is most effective in curbing the virus’ spread.

Vincent Feng, general director of ITRI’s Computational Intelligence Technology Center, says that ITRI is working closely with the government, using its technological expertise to fight the virus and staunch its spread. Its newly unveiled smart home-care technology can significantly reduce the burden on epidemic prevention personnel.

Also, he notes, ITRI will continue to employ its strengths in cross-disciplinary research to confront the far-reaching implications of COVID-19.

On the commercial side, one solution soon to go to market is the LivMote, a touchless temperature-screening device developed by Taipei-based hardware-as-a-service startup Soda Labs. The wall mounted device combines infrared thermography with a few other ambient sensors and engages machine learning to improve the accuracy of its temperature reads over time.

The LivMote, produced by Taipei-based startup Soda Labs, screens the temperatures of individuals at care facilities without the use of facial recognition technology. Photo: Soda Labs

Organizations that install the LivMote can choose what data it collects, whether it is stored in the device or in the cloud, and how long it is stored for. The security system, developed with Microsoft’s Azure IoT infrastructure, is built directly into the product.

Soda Labs CEO Andrew Jiang says that the LivMote is intended for use in senior living and long-term-care facilities, places that host particularly vulnerable populations. (At the end of April, the Wall Street Journal reported that the death toll from COVID-19 in American nursing homes had topped 10,000). For Jiang and his team, the goal in creating the product with long-term partner Foxconn was to devise something that helps safeguard the health of vulnerable segments of the populations, while also addressing privacy concerns in countries where facial recognition technology is seen as too invasive.

The LivMote was rolled out in just a matter of months. “We got to put to the test our ability to launch hardware/software products very quickly,” says Jiang. “We went from R&D concept in February to getting test units out to pilot locations in June, and we are moving on to mass production in August or September.”

Temperature-screening technology is a common thread among companies working on AI solutions to combat COVID-19. This trend is not surprising since almost 88% of confirmed cases of the virus have fevers, according to a February 2020 report from the World Health Organization.

Cardinal Tien Hospital in New Taipei City’s Yonghe District has collaborated with Microsoft Taiwan to install a device that screens the temperature of individuals entering the hospital and determines whether they are wearing the required face mask. Microsoft’s solution was developed on its Azure cloud-computing platform. The product, which was launched within two weeks, combines infrared sensors for temperature screening and AI to detect face-mask-wearing on an intelligent-edge device placed at the hospital’s entrance.

It employs the Azure Bot service to alert hospital authorities to any abnormalities in body temperature or visitors not wearing masks. In a Microsoft press release, Liao Mao-hung, Cardinal Tien Hospital’s Administration Vice Superintendent, noted that use of the device can reduce the workload of front-line medical personnel, saving on human resources.

Solutions are not limited to temperature screening. Veteran Taiwanese tech company Solomon Technology Corp., known for its AI-powered industrial automation products, has released two kinds of autonomous guided vehicle (AGV) robots. One is equipped with a UVC light and the other with a disinfectant atomizer. The products were developed with the help of Solomon’s Denmark-based partner, Mobile Industrial Robots.

According to Solomon Chairman Johnny Chen, the sterilizing AGVs can be deployed in settings such as hospitals, schools, and office buildings, helping ease the anxiety of people entering those buildings and reducing the burden on commercial cleaning services. Chen says that so far Solomon is focusing only on the Taiwan market for sales of the new AGVs.

Seizing opportunities

 COVID-19, while terrible, also presents tremendous opportunities for AI development and other innovative technologies worldwide. Taiwan is well positioned to take advantage of those opportunities, says Soda Lab’s Jiang.

“Taiwan has both the tech hardware manufacturing knowledge and the trust of the Western world,” Jiang says. It therefore has an opening now to lend its expertise and to build and improve products that will be used around the world. “I don’t think Taiwan necessarily has to push the boundaries of innovation in AI but can adopt international best practices and fit that together with hardware,” Jiang says.

Min Sun, Chief AI Scientist at digital marketing startup Appier, agrees, though he cautions that Taiwan and other countries need to continue to walk the line between developing AI solutions for good purposes like fighting COVID-19, while still protecting the personal data of their citizens. “I can see COVID-19 as a true growing force, not only in Taiwan, but also in countries where the situation is more severe. They need this more-efficient, data-driven, decision-making solution,” says Sun. “But we do need to be careful about the data privacy issue – to think carefully about how to achieve this state-of-the-art technology while also building systems that preserve privacy.”