WEBVTT 00:02.533 --> 00:05.066 align:left position:10% line:71% size:80% JUDY WOODRUFF: Even as more states are trying to reopen their economy, a new "PBS NewsHour"/NPR/Marist 00:07.066 --> 00:11.166 align:left position:10% line:77% size:80% poll found that 77 percent of Americans worry about a second wave of infections yet to come. 00:13.333 --> 00:18.333 align:left position:10% line:77% size:80% This comes as computer-based models suggest that the U.S. will pass its own grim milestone 00:20.333 --> 00:24.566 align:left position:10% line:77% size:80% by June, 100,000-plus deaths. That higher projection is arriving even sooner than some 00:26.900 --> 00:29.866 align:left position:20% line:83% size:70% of the models estimated just weeks ago. 00:29.866 --> 00:34.833 align:left position:10% line:77% size:80% But models are not crystal balls. The work that goes into making them and their ultimate 00:34.833 --> 00:39.800 align:left position:10% line:77% size:80% purpose is more complicated than you might be able to tell from the headlines. 00:41.866 --> 00:44.500 align:left position:10% line:77% size:80% Miles O'Brien explains in his latest report for our series the Leading Edge. 00:44.500 --> 00:49.466 align:left position:10% line:77% size:80% MILES O'BRIEN: We live in a complicated world, filled with more data than insight. 00:51.500 --> 00:56.166 align:left position:10% line:77% size:80% Finding a path to clarity is not easy, even on a good day. And these are not good days. 00:58.133 --> 01:03.100 align:left position:10% line:77% size:80% So, how can we take a huge amount of data and make it understandable, so we can see 01:04.233 --> 01:06.266 align:left position:30% line:89% size:60% the future? 01:06.266 --> 01:08.333 align:left position:10% line:77% size:80% BETZ HALLORAN, Fred Hutchinson Cancer Research Center: You can't believe every number that 01:08.333 --> 01:12.166 align:left position:10% line:77% size:80% comes out. But if we don't try to formulate our thinking about a complex process, then 01:13.033 --> 01:15.133 align:left position:10% line:89% size:80% we will be running blind. 01:15.133 --> 01:19.366 align:left position:10% line:77% size:80% MILES O'BRIEN: Betz Halloran is an infectious disease modeler. She writes mathematical formulas 01:20.733 --> 01:25.733 align:left position:10% line:83% size:80% that define the chaotic, exponential spread of infection. 01:27.666 --> 01:30.900 align:left position:10% line:77% size:80% A biostatistician at Seattle's Fred Hutchinson Cancer Research Center, she's part of the 01:32.800 --> 01:35.666 align:left position:10% line:77% size:80% team that curates the Global Epidemic and Mobility Model, or GLEAM. 01:35.666 --> 01:40.666 align:left position:10% line:77% size:80% BETZ HALLORAN: The GLEAM model is a big mobility model that can answer global questions. 01:42.633 --> 01:46.400 align:left position:10% line:77% size:80% MILES O'BRIEN: GLEAM begins with the first infection in China and travels down the many 01:48.800 --> 01:51.833 align:left position:10% line:71% size:80% paths of exponential growth, constantly calculating who is susceptible, exposed, infectious, and 01:54.666 --> 01:58.333 align:left position:10% line:89% size:80% recovered, S-E-I-R, or SEIR. 01:58.333 --> 02:02.166 align:left position:10% line:77% size:80% BETZ HALLORAN: You can structure it in many different ways. But, usually, when we talk 02:02.166 --> 02:06.900 align:left position:10% line:77% size:80% about infectious disease modeling, that's the basic sort of meat and potatoes of what's 02:06.900 --> 02:09.466 align:left position:20% line:89% size:70% going to be in a model. 02:09.466 --> 02:13.333 align:left position:10% line:71% size:80% MILES O'BRIEN: But the model does not stop there. It factors in the entire global transportation 02:15.000 --> 02:17.866 align:left position:10% line:83% size:80% network, including airline schedules and capacity. 02:17.866 --> 02:22.866 align:left position:20% line:71% size:70% BETZ HALLORAN: So, the question we were asking way back then was, where is it going to spread? 02:24.933 --> 02:27.800 align:left position:10% line:83% size:80% If it gets into the United States, where would it go first? 02:27.800 --> 02:32.800 align:left position:10% line:77% size:80% And once it gets in, then we could use GLEAM to look at the question of, how much is it 02:34.800 --> 02:37.133 align:left position:10% line:77% size:80% going to spread in the different places? Where is it going to go first? And then we predicted 02:37.133 --> 02:39.566 align:left position:20% line:89% size:70% that pretty well. 02:39.566 --> 02:43.133 align:left position:10% line:71% size:80% MILES O'BRIEN: Halloran and her team did accurately predict where COVID-19 would first surge in 02:43.933 --> 02:46.000 align:left position:20% line:89% size:70% the United States. 02:46.000 --> 02:50.200 align:left position:10% line:77% size:80% But, as the pandemic wore on, the limitations of the models became more evident. After all, 02:52.733 --> 02:56.466 align:left position:10% line:71% size:80% no one really knows how the virus is transmitted, who's likely to get sick and who won't, who's 02:57.666 --> 03:00.266 align:left position:20% line:83% size:70% likely to die, who might have immunity. 03:00.266 --> 03:04.500 align:left position:10% line:71% size:80% All those questions won't be answered until there is widespread testing. So, in the meantime, 03:06.366 --> 03:09.533 align:left position:10% line:77% size:80% the models muddle on, with sometimes dizzyingly confusing results. 03:11.500 --> 03:14.600 align:left position:10% line:77% size:80% One of them, from Britain's Imperial College, predicted two million COVID-19 deaths in the 03:14.600 --> 03:19.600 align:left position:10% line:77% size:80% United States. But that assumed no human response, no social distancing. 03:22.066 --> 03:24.500 align:left position:10% line:71% size:80% BETZ HALLORAN: All models are wrong, but some models are helpful, and I think it's important 03:24.500 --> 03:26.533 align:left position:20% line:89% size:70% to remember that. 03:26.533 --> 03:30.033 align:left position:10% line:77% size:80% MILES O'BRIEN: Nearby, at the University of Washington's Institute for Health Metrics 03:30.033 --> 03:35.033 align:left position:10% line:77% size:80% and Evaluation, they built a much simpler model that started with a specific question 03:37.500 --> 03:40.633 align:left position:10% line:71% size:80% in mind: Did the health care system have the capacity to treat a surge of COVID-19 patients? 03:43.666 --> 03:48.666 align:left position:10% line:77% size:80% Chris Murray is the director. He and his team wrote a model that, unlike many others at 03:50.033 --> 03:52.433 align:left position:10% line:83% size:80% the time, factored in the human response to the pandemic. 03:52.433 --> 03:54.133 align:left position:10% line:77% size:80% DR. CHRISTOPHER MURRAY, Director of Health Metrics, University Of Washington: If you 03:54.133 --> 03:57.966 align:left position:20% line:77% size:70% ignore the behavioral response, you're going to massively overshoot. 03:57.966 --> 04:02.933 align:left position:20% line:71% size:70% And so I think it is a reasonable strategy to try to look at models, like the economists 04:05.233 --> 04:09.933 align:left position:10% line:77% size:80% do, which build in how individuals, local government, state government, are going to 04:09.933 --> 04:12.466 align:left position:20% line:83% size:70% respond to the problems as they unfold. 04:12.466 --> 04:13.700 align:left position:10% line:71% size:80% DR. DEBORAH BIRX, White House Coronavirus Response Coordinator: So I'm sure you're interested 04:13.700 --> 04:15.766 align:left position:10% line:89% size:80% in seeing all the states. 04:15.766 --> 04:20.033 align:left position:10% line:77% size:80% MILES O'BRIEN: Producing speedy state-by-state results, with consistently lower projections, 04:22.366 --> 04:25.566 align:left position:10% line:71% size:80% the University of Washington model was frequently cited by the White House in daily coronavirus 04:26.333 --> 04:28.066 align:left position:30% line:89% size:60% briefings. 04:28.066 --> 04:30.100 align:left position:20% line:77% size:70% DR. DEBORAH BIRX: And I think, if you ask Chris Murray, he would say... 04:30.100 --> 04:33.800 align:left position:10% line:77% size:80% MILES O'BRIEN: But the model initially assumed there would be widespread adoption of social 04:33.800 --> 04:37.733 align:left position:20% line:83% size:70% distancing restrictions in the U.S. 04:37.733 --> 04:42.533 align:left position:10% line:77% size:80% Once it became clear that wasn't happening, the modeling team went back to the drawing 04:42.533 --> 04:47.533 align:left position:10% line:77% size:80% board, releasing a new version on May 4. It now uses mobility data gleaned from cell phone 04:49.433 --> 04:54.333 align:left position:10% line:77% size:80% usage to better understand how well people are complying with the expert advice. 04:56.766 --> 05:00.300 align:left position:10% line:71% size:80% As a result, that model's projection for the total U.S. death toll by August 4 from COVID-19 05:02.100 --> 05:04.800 align:left position:20% line:83% size:70% instantly went from about 72,000 to 134,000. 05:04.800 --> 05:09.800 align:left position:10% line:77% size:80% DR. CHRISTOPHER MURRAY: It's sensible to try to look at a wide array of models and try 05:11.866 --> 05:16.866 align:left position:10% line:77% size:80% to look at how -- do they tell you the same story? Are they converging? 05:19.100 --> 05:22.066 align:left position:10% line:71% size:80% It's very confusing, I think, for many decision-makers to navigate through some of the models. 05:24.466 --> 05:26.500 align:left position:10% line:83% size:80% MAN: We're going to start off with this weekend. 05:26.500 --> 05:31.266 align:left position:20% line:71% size:70% MILES O'BRIEN: Weather forecasters are some of the most adept at navigating the inherent 05:31.266 --> 05:32.866 align:left position:10% line:89% size:80% uncertainties of modeling. 05:32.866 --> 05:35.000 align:left position:20% line:83% size:70% MAN: Going to have some travel problems if... 05:35.000 --> 05:39.333 align:left position:10% line:77% size:80% MILES O'BRIEN: After all, it's been 70 years since they first ran a model through a computer 05:41.333 --> 05:44.500 align:left position:10% line:77% size:80% to create a forecast. It's been steady improvement ever since. It's now possible to reliably 05:45.900 --> 05:50.433 align:left position:10% line:83% size:80% forecast seven days in advance with 80 percent accuracy. 05:50.433 --> 05:55.433 align:left position:10% line:77% size:80% But, with a novel virus, there are so many unknowns. And weather models do not have to 05:56.900 --> 06:00.466 align:left position:10% line:89% size:80% account for human behavior. 06:00.466 --> 06:05.100 align:left position:10% line:77% size:80% Marshall Shepherd is director of the Atmospheric Sciences Program at the University of Georgia. 06:05.100 --> 06:07.133 align:left position:10% line:77% size:80% MARSHALL SHEPHERD, Atmospheric Sciences Program Director, University of Georgia: It's very 06:07.133 --> 06:11.700 align:left position:10% line:77% size:80% important, when consuming these coronavirus models and weather models, to consume the 06:11.700 --> 06:13.733 align:left position:20% line:83% size:70% uncertainty that we know is inherent. 06:13.733 --> 06:16.966 align:left position:10% line:77% size:80% But we have a way to get around that in weather called ensemble modeling. 06:16.966 --> 06:21.866 align:left position:10% line:77% size:80% MILES O'BRIEN: Ensemble modeling, meaning combining the predictions of many different 06:21.866 --> 06:26.866 align:left position:10% line:77% size:80% models, it's a crucial tool that has greatly improved forecasting the weather and, in the 06:28.200 --> 06:32.233 align:left position:10% line:83% size:80% past three years, seasonal influenza as well. 06:32.233 --> 06:37.233 align:left position:10% line:71% size:80% Nick Reich is an associate professor of biostatistics at the University of Massachusetts-Amherst. 06:39.166 --> 06:43.500 align:left position:10% line:77% size:80% Working with the Centers for Disease Control and Prevention, he leads a team that builds 06:43.500 --> 06:47.766 align:left position:20% line:77% size:70% ensemble models to improve predictions of the spread of the flu. 06:47.766 --> 06:50.933 align:left position:10% line:71% size:80% NICK REICH, University of Massachusetts-Amherst: I don't think any one model should be viewed 06:50.933 --> 06:53.000 align:left position:30% line:89% size:60% as gospel truth. 06:53.000 --> 06:57.666 align:left position:10% line:77% size:80% When you just use one model, you end up with a too strong reliance on one particular set 07:00.200 --> 07:04.166 align:left position:10% line:71% size:80% of assumptions and one particular viewpoint. And this is why it's really critical to consider 07:06.266 --> 07:08.733 align:left position:10% line:89% size:80% multiple models together. 07:08.733 --> 07:13.633 align:left position:10% line:71% size:80% MILES O'BRIEN: The influenza models are informed by up to 20 years of experience with the viruses 07:15.300 --> 07:18.066 align:left position:10% line:89% size:80% and the accuracy of the models. 07:18.066 --> 07:23.066 align:left position:10% line:77% size:80% Reich and his team have now built a COVID-19 ensemble model. But it, of course, does not 07:24.233 --> 07:26.233 align:left position:20% line:83% size:70% have the benefits of a long backstory. 07:26.233 --> 07:31.100 align:left position:10% line:77% size:80% NICK REICH: We do have hundreds of years of theory about how to build mathematical models 07:33.433 --> 07:38.433 align:left position:10% line:77% size:80% of infectious disease, but have they ever been tested in real time in this way, with 07:40.766 --> 07:44.166 align:left position:10% line:83% size:80% all of the data sources that are available to us? No. 07:44.166 --> 07:48.833 align:left position:10% line:77% size:80% We're building this car as it's careening down the highway, and we're learning about 07:48.833 --> 07:51.066 align:left position:20% line:89% size:70% these models as we go. 07:51.066 --> 07:55.433 align:left position:10% line:77% size:80% MILES O'BRIEN: Infectious disease modelers are scrambling to figure out where we are 07:55.433 --> 07:58.900 align:left position:20% line:83% size:70% headed, depending on the decisions we make. 07:58.900 --> 08:03.900 align:left position:10% line:77% size:80% If we take the time to better understand what the models can and cannot do, maybe we will 08:05.333 --> 08:09.500 align:left position:10% line:83% size:80% do the same as we search for the path back to normalcy. 08:10.600 --> 08:12.166 align:left position:20% line:83% size:70% For the "PBS NewsHour," I'm Miles O'Brien.