April 01, 2020
JERUSALEM — Teams of epidemiologists and computer scientists on three continents have started mass population surveys to try to get ahead of the coronavirus and ensure that scarce diagnostic tests, and even scarcer ventilators, are sent where they can do the most good.
More than two million people in Britain and 150,000 Israelis have already completed simple questionnaires, and many are updating their answers daily. Analysts of the data — including symptoms of Covid-19 and test results, as well as risk factors and demographics — say they have been able to identify incipient outbreaks days ahead of the authorities.
Three groups in the United States — led by Massachusetts General Hospital, the Massachusetts Institute of Technology and Weill Cornell Medicine in New York — are now vying to attract enough survey participants nationwide to detect impending hot spots.
In Israel, where artificial-intelligence experts at the Weizmann Institute of Science have played a leading role in the effort, computer models that were fed the results of questionnaires accurately predicted surges in cases in cities like Bnei Brak, Jerusalem and Beersheva five days in advance, researchers said.
The questionnaires, which are web or app based, ask people about their medical histories and risk factors, and then about symptoms. In Israel, a bot prompts users to choose from a menu including coughing, sore throat, shortness of breath, fatigue and a loss of taste or smell, and then asks for a temperature reading from the last 24 hours.
“Imagine, if the Ministry of Health had acted on that information, maybe they could have saved more spreading of the virus,” said Eran Segal, a computational biologist at Weizmann who is spearheading the project.
Forecasting areas of infection would allow officials to flood those places with testing capacity, Dr. Segal said. And it would allow doctors to better marshal their scant resources.
“Every time they test somebody and it’s negative, it’s a wasted test,” he said. “You want to get positives, act on it, and isolate people.”
The Israeli survey, in six languages, began by word of mouth, Dr. Segal said. But this week, Israel’s largest H.M.O., Clalit, sent text messages to its 4.5 million subscribers pleading with them to participate.
“Help save your loved ones and yourself,” it said.
In Britain, researchers said the disease had advanced so fast that it was already pointless to use survey results to steer tests for the virus; rather, the results are helping pinpoint where ventilators and mobile intensive-care units should be positioned, according to Tim Spector, a professor of genetic epidemiology at King’s College, London.
“I liken this to a radar,” he said. “Two weeks before the bombs get dropped, you can actually work out what’s going to happen — not based on what happened in China, but actually seeing what’s happening on the ground here. That’s what’s going to save lives: the fact that you can plan where to put your mobile I.C.U.s.”
Dr. Spector, the director of the TwinsUK Registry, a nationwide research cohort of 15,000 twins, rolled out the British survey to that group on March 24, and it spread by word of mouth. Within five days, he said, the tally of unique participants hadsurpassed two million, with the “Covid Symptom Tracker” app jumping to the No. 1 health-related download in Britain's Apple Store and the No. 4 app of any kind.
The survey is popular, but not as evenly distributed as health experts might like. People over 70 years old are underrepresented, which is not a surprise, given the app-based distribution. And women are participating at roughly twice the rate of men. “They are more altruistic, in the U.K.,” Dr. Spector said, laughing.
Over all, he said, the respondents in Britain are acting out of concern for the National Health Service and “a feeling that they’re doing something for the community.” But he also said that people who reported already having tested positive for the virus seemed to grasp at the survey as “a way to express their symptoms.”
“If they don’t go to hospital, nobody seems to care about them,” Dr. Spector said.
Surveys are not the only unconventional way that scientists are trying to detect the spread of the virus: A maker of smart thermometers says it is tracking the contagion in real time by mapping fevers in American households.
In Britain, the survey suggested that loss of taste and smell could be more predictive of a positive test result than self-reported fever, fatigue or loss of appetite, Dr. Spector said. That could be of critical importance to health workers who may still be going to work despite experiencing a loss of taste or smell, but isolating themselves based on other symptoms, he said.
But he emphasized that much was still to be learned.
“This data’s still wet,” Dr. Spector said. “We have teams of people working on the analysis to try and get stuff out that’s going to be of critical importance for the health service.”
Unlike countries with national health systems, the United States, with its atomized health industry, poses a difficult challenge in reaching enough respondents to track the epidemic at the local level.
“So we need to develop creative solutions,” said Andrew Chan, an epidemiologist at Massachusetts General and professor of immunology and infectious diseases at Harvard.
Dr. Chan’s team, which is adapting the British app for the United States, has joined with Stand Up to Cancer, tapping its list of 1.5 million cancer patients and their relatives. It is also drawing on a research cohort of about 51,000 female nurses across the country.
Starting with those audiences, Dr. Chan said, offers added benefits: Both cancer patients and health workers are at particular risk.
In the M.I.T.-led project, researchers are preparing to promote a survey app that they developed with the help of engineers from Pinterest, the image-sharing site, whose co-founder, Ben Silbermann, was a high school classmate in Des Moines of Feng Zhang, a biochemist leading the M.I.T. survey effort.
Dr. Zhang said that he and his colleagues hit on the idea for the surveys just after they began working from home a few weeks ago. One of them, who had been a student of Dr. Segal’s in Israel, saw mention of the Israeli survey project, and the two groups began to coordinate.
Dr. Zhang and Dr. Segal are already taking their project to other corners of the globe: A white paper they drafted has co-authors in Argentina, Canada, Estonia, Germany, Luxembourg, Slovenia, Sweden and Switzerland.
One area of difference among the surveys concerns privacy.
The British survey allows people to volunteer personal information, and Dr. Spector said it might become possible to tie in the survey results to the National Health Service’s database. The Massachusetts General project allows cancer patients in clinical trials to provide contact information for their study physicians if they choose, “to let them know if they are developing symptoms,” Dr. Chan said.
But the Israeli questionnaires ask only the name of a respondent’s street, not the house number. And Olivier Elemento, director of Weill Cornell’s Englander Institute for Precision Medicine in New York, said its own survey, which includes a dashboard so respondents can see how the data is used, was deliberately collecting only county-level information.
“We want people to feel they can participate without any fear of being identified,” he said.
The researchers acknowledge without hesitation that surveys completed by laypeople without corroboration by scientists or physicians will lack the precision of more methodical research. But they say widespread participation will produce enough data to largely overcome the effects of personal variations.
“Given the speed of the pandemic, we don’t have the luxury of waiting to validate every report we get in from a participant,” said Dr. Chan, the epidemiologist at Massachusetts General. “We need to act quickly.”
David D. Celentano, chairman of epidemiology at the Johns Hopkins Bloomberg School of Public Health, praised the efforts to gather data. “In the U.S., where we had so little testing till last week, we’ve been completely blind,” he said. “We have no idea where the epidemic is.”
Dr. Celentano cautioned that “people who are going to pay attention to their phone and do this every day are probably either the worried well or people who are experiencing symptoms, so you’ll have a very biased sample.”
Still, he said, however flawed, the surveys could be “very useful for targeting where to put health services.”
Or, as Mr. Segal put it: “The rule of big numbers means we don’t need to be super-super-accurate. We just need to be able to rank this region over that region.”
While they are racing now to overtake the coronavirus, the researchers say their survey tools could, alas, come in handy again.
“This has been a lesson in the limitations of our ability to respond as epidemiologists,” Dr. Chan said of the coronavirus. “We need to develop tools that will allow us in the future to respond more quickly to data collection and analysis in real time. Because this may not be the last pandemic.”
# # #