The Doctor Whose Gut Instinct Beat AI in Spotting the Coronavirus

The New York epidemiologist who alerted the world to COVID-19 outbreak.

Dr. Margorie Pollack

Information is critical to stemming the global spread of the new coronavirus, and few people have been as effective as Dr. Marjorie Pollack and her colleagues at ProMED. This largely volunteer program of the International Society for Infectious Diseases sent the first detailed alert about the appearance of a new pneumonia-like outbreak in Wuhan, China, just as it was the first to report the outbreak of Severe Acute Respiratory Syndrome, or SARS, back in 2003. As of March 5, the virus, SARS-CoV-2, which causes the disease COVID-19, has claimed more than 3,300 lives, mostly in China but also significant numbers in South Korea, Iran, and Italy, and 11 in the United States.

On Dec. 30, Pollack was checking emails after dinner when she got an alert from a colleague about clusters of seriously ill patients in China. She immediately went to work contacting other ProMED colleagues to verify what was happening. Four hours later, an artificial intelligence system run by Boston Children’s Hospital sent out a brief alert about unidentified pneumonia cases in Wuhan, and rated its seriousness as a 3 out of 5. Half an hour later, just shy of midnight, Pollack sent a more-detailed and useful warning to ProMED’s global community of some 80,000 doctors, epidemiologists, public health officials, and other experts and interested lay persons. She said her experience with SARS nearly two decades ago helped her and colleagues spot the dangerous new threat despite all the social media noise coming out of China.

Pollack, a New York-based physician epidemiologist who has worked in more than 50 countries for the US Centers for Disease Control and Prevention, the World Health Organization, and other public health bodies, is encouraged by China’s apparent success in slowing the spread of the disease. But she says there is still much we do not know, including how prevalent the virus is in the population and its true lethality. She spoke by telephone with Partha Bose, a partner at Oliver Wyman and a leader of the Oliver Wyman Forum, as well as Jilian Mincer, managing editor of the Oliver Wyman Forum, and Tom Buerkle, the Forum’s online editor.

This interview was conducted on February 28. The World Health Organization declared the coronavirus outbreak to be a pandemic on March 11.

Oliver Wyman Forum: How were you and your colleagues able to call attention to the coronavirus outbreak, COVID-19, so quickly?

Marjorie Pollack: I got an alert from a colleague who keeps a finger on the pulse of Weibo, the Chinese social media platform. The alert gave me some tweets about stuff that was going on in Wuhan – a cluster of four cases, then 27 cases - along with a picture theoretically of a document sent out by the Wuhan public health commission stating something about pneumonia cases that seemed to be associated with a seafood and wildlife market. Having lived through and worked through the SARS outbreak, it just rang a bell. This was a déjà vu.

Did HealthMap, the AI system at Boston Children’s Hospital, play a role in this?

We collaborate with HealthMap. We do use their information. In 2003, the alert we got on SARS was truly a rumor. It was a physician subscriber who said he heard from a friend who has a friend who belongs to a teacher’s chatroom in Guangdong, China, reporting that hospitals are filled, people are dying everywhere, massive panic.

COVID was a similar process. We didn’t know if the picture in the Weibo post was real. So we tried looking around and found a media report saying the Wuhan health commission validated that the document was real. So we sent out an alert the evening of Dec. 30. On the 31st, China reported it to WHO officially. Which is interesting. If you go back to SARS, we put an alert out Feb. 10, and China officially reported it Feb. 11.

Do you have reason to believe your information prompted them?

I think it helped. I think this go around, China has been completely transparent. SARS was a lesson on the need for transparency. I’ve been very impressed with them. They’ve been putting out data whenever they’ve had it. I think in Wuhan, what happened was they were just overwhelmed. And they were very honest. They admitted that they basically didn’t have the surge capacity to handle the volume, which is why they ended up building two hospitals in less than a week.

What are the advantages and disadvantages of your community networking approach compared with AI?

It’s, what can I say, a gut reaction. It’s still the human factor. The computers still haven’t totally replaced the humans.

With the AI outputs at present, anywhere you look, there’s information overload. What we are looking for can be the equivalent of a needle in a haystack. Those early reports coming out of China for the first 24 hours were the needle in the haystack - the signal, if you will. When we review media reports or go through the literally hundreds of emails and social media posts we get every day, we deal with a signal-to-noise ratio. Our signal is that needle in a haystack. The astute clinician, the astute practitioner, has been the one tipping what’s going on – they have been able to cut through the noise and call out the signal -- in this case a new disease caused by a new virus not previously seen in humans.

Are we ahead of the curve on COVID-19 because of what we learned from SARS?

It’s not yet completely clear. We’re ahead of the curve in a lot of ways but we’re only two months into experience with this new coronavirus.

Wuhan happens to have an excellent virology laboratory. And they’ve been doing research on coronaviruses since SARS. The new virus was identified in just about two weeks. They used nuclear acid testing to find the genome. With SARS, we didn’t have this technology. People had to culture. It took three months to identify the virus strain.

How did you get involved with ProMED?

ProMED began in 1994 as an attempt to see if the then nascent Internet could be used as a source of communication. I've always been interested in disease surveillance and novel sources of information on diseases and outbreaks.  I learned of the ProMED initiative through word of mouth; one of the founders, Dr. Jack Woodall, had been a CDC colleague.  It was a natural fit given the matching interests.  We used what we would consider nontraditional sources for early warning. What we found was the media was an excellent source. There would be “rumors” that a village or town was having some sort of an outbreak. It didn’t come to the attention of the national surveillance system, but it came through the media – radio and the press, mostly.

Are your subscribers mostly epidemiologists?

No, it’s very mixed. We use what’s called the One Health model. It basically recognizes that humans, animals, plants, and the environment all interact, and you need to keep your finger on the pulse of all aspects of what’s going on.

Of "new diseases" in humans, 75 percent of those identified in the past 25 to 50 years have been species jumps from domestic animals, including livestock.  And 70 percent of new diseases in domestic animals are jumps from wildlife.  COVID-19 was a species jump, most likely from bats to an as yet unidentified intermediary host, and then to humans.

Do you think China’s stance that the rate of contagion is going down is accurate?

Yes, with one caveat. We still don’t have a handle on what the true infection rate is. We’ve been waiting to get a good serology test and do a survey to see how many people have antibodies to this virus. I suspect we’re going to find it’s a much less severe disease overall, that the case fatality rate will be less than 1 percent. What you see with most new diseases, your initial picture are the very severe cases that come to the attention of the health sector.

Is it time to declare this a pandemic?

I don’t think this is a pandemic yet – though we are still in the early stages of finding out more about its characteristics and reach. My favorite adage is ‘seek and ye shall find.’ And the corollary is, ‘don’t look and it ain’t there.’ And I think that’s some of what we are seeing – we don’t have an accurate picture of how far it has spread.

I think it’s a virus that has developed very easy person-to-person transmission. I am hoping that it is going to end up being predominately a common cold virus, with just some high-risk individuals having much more serious disease. Some of it is a function of age, some of it is a function of pre-existing co-morbidities, and some of it may have a genetic predisposition.

What should people be doing?

The general recommendations are the same recommendations for any of the seasonal flus: Frequent hand washing, because a lot of the transmission is from your hands. You don’t want to eat or rub your face if you’ve been touching a lot of things around you. And try to maintain a distance from others and try to avoid crowded gatherings. The same kind of behavior you would do when we have a new flu virus coming around.

Is there anything you would advise cities or local governments to be doing?

Epidemic and pandemic preparedness has worked on increasing surge capacity. Theoretically cities have pandemic plans. The question is, will we go the route other countries have gone to try to slow transmission by limiting mass gatherings. Japan just closed down schools for the next month. And they’re asking parents to keep children at home. China has a culture where people who come down with even a common cold try to use a mask to prevent transmission. We might not have similar cultural traits or be able to take similar actions everywhere around the world.

Moreover, reducing gatherings has significant economic implications. So, in summary, it’s a tough call.