▲ Ching-Yun Lin, the founder of Graphen, has been travelling to and from Taiwan to promote AI applications in recent years. (Photo / Photography Group)
On the way to the office on Fifth Avenue in Manhattan, Ching-Yun Lin, the founder of Graphen, noticed that the crowd has suddenly become sparse. "Starting today, I also work from home." Ching-Yun Lin said with a bitter smile in a telephone interview.
He is a well-known AI technology expert and formerly the Chief Scientist at IBM. Recently, he and the company team were busy using AI technology to study the Coronavirus genes in Wuhan and track down how the "most deceptive virus" spread to the world.
Ching-Yun Lin's computer stores genetic data of hundreds of virus variants of the Coronavirus from 30 countries. On March 10, he and his team compared 30,000 genes in each virus and finally completed the world's first Coronavirus spreading map. This is also the first time that the world can clearly see how the virus that endangers the world has escaped from China in just over 3 months and has also evolved a subtype virus that has hit the world.
AI combined with public health breakthrough to seize genetic mutation
This is a major breakthrough in combining AI with public health. In the previous analysis, we could only assume that every patient got the same virus, and speculate on the future development of the epidemic. "This is the first time that we can see the whole and "You can see the changes of each virus." Just by injecting more information, you can use such a system to further predict the future development of the epidemic.
Ching-Yun Lin explained that most people regard Coronavirus as a virus. In fact, every time a new infected person is transmitted, the gene of the virus may be changed randomly. "Recently I heard of a new virus in Beijing that attacks the central nervous system. worry". He analyzed that the more people infected, the more times the virus mutated, and it was hard to guarantee that no more aggressive new viruses would emerge.
Just as police officers use fingerprints to catch thieves, Ching-Yun Lin uses AI as a magnifying glass to compare the "genetic fingerprints" of each virus. In a week, he finds the epidemic prevention breaches that governments will take months to find. For example, the United States only began to be aware of the increased epidemic in early March, but Ching-Yun Lin's analysis found that Coronavirus spreads in three ways, and two key super communicators had planted the bane of the European and American pandemic as early as January.
The first super-spreader was a Wuhan patient discovered on January 5. The variant gene carried on him was later found in Taiwan, Belgium, and Australia, and finally landed in Washington State in the United States. Washington State secretly spread it for a month, "and then to the Supreme Princess cruise ship in California, and to Minnesota in early March.
Another super communicator is the source of the European pandemic. Ching-Yun Lin discovered that a patient diagnosed on January 24 had a viral gene spread to Germany, Italy and Switzerland, which eventually caused a pandemic in Europe. He observed that the viral genes that are prevalent in Europe already carry components different from those in China. His analysis pointed out that Australia was another major source of virus invasion in Europe. The same virus spread in Singapore, France, and Italy, and stepped towards Europe step by step like a springboard.
The third route spreads in Asia. Infected patients on the cruise ships of Beijing, South Korea and Diamond Princess, most of the virus genes they infected are very similar to the original virus in Wuhan.
▲ Coronavirus spread map produced by Ching-Yun Lin and his team, indicating the relationship between the path of virus spread and infected cases. (Picture / taken from Graphen)
Find out the path of transmission
Ching-Yun Lin pointed out that spreading out this map found that Taiwan's fight against Coronavirus is not enough. He used the Taiwan case he obtained to show that someone had traveled to the Middle East to be infected with the virus, and it was found out that he was carrying Coronavirus from Europe.
The situation in the United States is not clear. Like a few days ago, a lawyer in New York City was found to have infected more than 100 people. "He hasn't been abroad, and he hasn't been able to find where he was infected." Ching-Yun Lin said that there may be undiscovered sources of infection in the United States.
As for Taiwan, Ching-Yun Lin observed that it must resist multiple sources of infection. "In seven cases, the virus genes were the same as the original genes in Wuhan, one from Europe and two from Asia." The number of cases is not large, and it is not easy to withstand so many attacks.
He observed that the outbreaks in Japan and China had gradually been brought under control. "The local virus gene in Japan has not spread." He explained that few other viruses with the same genome as Japanese patients were found in other places, indicating that Japanese control has been effective, fewer new cases in China, and Singapore has control. Living in Europe, the epidemic prevention was not good before, I am afraid it is a new threat to epidemic prevention. These data can help countries to identify where the threat of the epidemic comes from or to develop more effective drugs.
From finding contacts, counterterrorism to epidemic prevention, AI analysis is widely used
Interestingly, he was able to complete the analysis so quickly because he had presided over the US government's plan to counter terrorism. Ching-Yun Lin was originally an expert in Internet research. In April 2009, the US Business Week reported his innovative research "How to Put a Price on Social Connection". Ching-Yun Lin analyzed everyone's interaction on the Internet I found that a good connection can bring in $ 986 in earnings per month!
"Seeing this cover, the US government came to me," said Ching-Yun Lin. The US government wants to know how to find terrorists by analyzing the Internet.
Ching-Yun Lin won two big projects, with scholars from IBM and 10 American universities, to develop a system that can analyze huge amounts of information, and to find secret organizations and gold streams that are invisible to ordinary people on the Internet.
A Taiwanese AI scholar observed that Ching-Yun Lin's research is characterized by "his system can handle extremely large amounts of data." In the past, he used such a technique to analyze the Internet around the world to find terrorists, but now he just changes his target and uses genetic data to analyze the spreading virus. At present, Graphen's AI technology has cooperated with many institutions to use AI for bank anti-money laundering.
Ching-Yun Lin's next step is to promote AI-driven precision medicine. "We cooperate with the Chinese Academy of Research and use AI to analyze protein mutations and predict changes in cancer genes." The principle and analysis of epidemic conditions are a bit like that. It can help analyze the trend of cancer cell changes and "determine which drug is more effective." At present, a New York company is also working with them to use big data medical records to predict general conditions.
Maybe when the next epidemic strikes, we will not only rely on masks for protection, but also let the AI system from Taiwan predict for us the risk of infection that every country, even every passenger flying into Taiwan, may bring.
Source: Original story in Chinese https://www.wealth.com.tw/home/articles/24754