1 00:00:02,533 --> 00:00:05,066 JUDY WOODRUFF: Even as more states are trying to reopen their economy, a new "PBS NewsHour"/NPR/Marist 2 00:00:07,066 --> 00:00:11,166 poll found that 77 percent of Americans worry about a second wave of infections yet to come. 3 00:00:13,333 --> 00:00:18,333 This comes as computer-based models suggest that the U.S. will pass its own grim milestone 4 00:00:20,333 --> 00:00:24,566 by June, 100,000-plus deaths. That higher projection is arriving even sooner than some 5 00:00:26,900 --> 00:00:29,866 of the models estimated just weeks ago. 6 00:00:29,866 --> 00:00:34,833 But models are not crystal balls. The work that goes into making them and their ultimate 7 00:00:34,833 --> 00:00:39,800 purpose is more complicated than you might be able to tell from the headlines. 8 00:00:41,866 --> 00:00:44,500 Miles O'Brien explains in his latest report for our series the Leading Edge. 9 00:00:44,500 --> 00:00:49,466 MILES O'BRIEN: We live in a complicated world, filled with more data than insight. 10 00:00:51,500 --> 00:00:56,166 Finding a path to clarity is not easy, even on a good day. And these are not good days. 11 00:00:58,133 --> 00:01:03,100 So, how can we take a huge amount of data and make it understandable, so we can see 12 00:01:04,233 --> 00:01:06,266 the future? 13 00:01:06,266 --> 00:01:08,333 BETZ HALLORAN, Fred Hutchinson Cancer Research Center: You can't believe every number that 14 00:01:08,333 --> 00:01:12,166 comes out. But if we don't try to formulate our thinking about a complex process, then 15 00:01:13,033 --> 00:01:15,133 we will be running blind. 16 00:01:15,133 --> 00:01:19,366 MILES O'BRIEN: Betz Halloran is an infectious disease modeler. She writes mathematical formulas 17 00:01:20,733 --> 00:01:25,733 that define the chaotic, exponential spread of infection. 18 00:01:27,666 --> 00:01:30,900 A biostatistician at Seattle's Fred Hutchinson Cancer Research Center, she's part of the 19 00:01:32,800 --> 00:01:35,666 team that curates the Global Epidemic and Mobility Model, or GLEAM. 20 00:01:35,666 --> 00:01:40,666 BETZ HALLORAN: The GLEAM model is a big mobility model that can answer global questions. 21 00:01:42,633 --> 00:01:46,400 MILES O'BRIEN: GLEAM begins with the first infection in China and travels down the many 22 00:01:48,800 --> 00:01:51,833 paths of exponential growth, constantly calculating who is susceptible, exposed, infectious, and 23 00:01:54,666 --> 00:01:58,333 recovered, S-E-I-R, or SEIR. 24 00:01:58,333 --> 00:02:02,166 BETZ HALLORAN: You can structure it in many different ways. But, usually, when we talk 25 00:02:02,166 --> 00:02:06,900 about infectious disease modeling, that's the basic sort of meat and potatoes of what's 26 00:02:06,900 --> 00:02:09,466 going to be in a model. 27 00:02:09,466 --> 00:02:13,333 MILES O'BRIEN: But the model does not stop there. It factors in the entire global transportation 28 00:02:15,000 --> 00:02:17,866 network, including airline schedules and capacity. 29 00:02:17,866 --> 00:02:22,866 BETZ HALLORAN: So, the question we were asking way back then was, where is it going to spread? 30 00:02:24,933 --> 00:02:27,800 If it gets into the United States, where would it go first? 31 00:02:27,800 --> 00:02:32,800 And once it gets in, then we could use GLEAM to look at the question of, how much is it 32 00:02:34,800 --> 00:02:37,133 going to spread in the different places? Where is it going to go first? And then we predicted 33 00:02:37,133 --> 00:02:39,566 that pretty well. 34 00:02:39,566 --> 00:02:43,133 MILES O'BRIEN: Halloran and her team did accurately predict where COVID-19 would first surge in 35 00:02:43,933 --> 00:02:46,000 the United States. 36 00:02:46,000 --> 00:02:50,200 But, as the pandemic wore on, the limitations of the models became more evident. After all, 37 00:02:52,733 --> 00:02:56,466 no one really knows how the virus is transmitted, who's likely to get sick and who won't, who's 38 00:02:57,666 --> 00:03:00,266 likely to die, who might have immunity. 39 00:03:00,266 --> 00:03:04,500 All those questions won't be answered until there is widespread testing. So, in the meantime, 40 00:03:06,366 --> 00:03:09,533 the models muddle on, with sometimes dizzyingly confusing results. 41 00:03:11,500 --> 00:03:14,600 One of them, from Britain's Imperial College, predicted two million COVID-19 deaths in the 42 00:03:14,600 --> 00:03:19,600 United States. But that assumed no human response, no social distancing. 43 00:03:22,066 --> 00:03:24,500 BETZ HALLORAN: All models are wrong, but some models are helpful, and I think it's important 44 00:03:24,500 --> 00:03:26,533 to remember that. 45 00:03:26,533 --> 00:03:30,033 MILES O'BRIEN: Nearby, at the University of Washington's Institute for Health Metrics 46 00:03:30,033 --> 00:03:35,033 and Evaluation, they built a much simpler model that started with a specific question 47 00:03:37,500 --> 00:03:40,633 in mind: Did the health care system have the capacity to treat a surge of COVID-19 patients? 48 00:03:43,666 --> 00:03:48,666 Chris Murray is the director. He and his team wrote a model that, unlike many others at 49 00:03:50,033 --> 00:03:52,433 the time, factored in the human response to the pandemic. 50 00:03:52,433 --> 00:03:54,133 DR. CHRISTOPHER MURRAY, Director of Health Metrics, University Of Washington: If you 51 00:03:54,133 --> 00:03:57,966 ignore the behavioral response, you're going to massively overshoot. 52 00:03:57,966 --> 00:04:02,933 And so I think it is a reasonable strategy to try to look at models, like the economists 53 00:04:05,233 --> 00:04:09,933 do, which build in how individuals, local government, state government, are going to 54 00:04:09,933 --> 00:04:12,466 respond to the problems as they unfold. 55 00:04:12,466 --> 00:04:13,700 DR. DEBORAH BIRX, White House Coronavirus Response Coordinator: So I'm sure you're interested 56 00:04:13,700 --> 00:04:15,766 in seeing all the states. 57 00:04:15,766 --> 00:04:20,033 MILES O'BRIEN: Producing speedy state-by-state results, with consistently lower projections, 58 00:04:22,366 --> 00:04:25,566 the University of Washington model was frequently cited by the White House in daily coronavirus 59 00:04:26,333 --> 00:04:28,066 briefings. 60 00:04:28,066 --> 00:04:30,100 DR. DEBORAH BIRX: And I think, if you ask Chris Murray, he would say... 61 00:04:30,100 --> 00:04:33,800 MILES O'BRIEN: But the model initially assumed there would be widespread adoption of social 62 00:04:33,800 --> 00:04:37,733 distancing restrictions in the U.S. 63 00:04:37,733 --> 00:04:42,533 Once it became clear that wasn't happening, the modeling team went back to the drawing 64 00:04:42,533 --> 00:04:47,533 board, releasing a new version on May 4. It now uses mobility data gleaned from cell phone 65 00:04:49,433 --> 00:04:54,333 usage to better understand how well people are complying with the expert advice. 66 00:04:56,766 --> 00:05:00,300 As a result, that model's projection for the total U.S. death toll by August 4 from COVID-19 67 00:05:02,100 --> 00:05:04,800 instantly went from about 72,000 to 134,000. 68 00:05:04,800 --> 00:05:09,800 DR. CHRISTOPHER MURRAY: It's sensible to try to look at a wide array of models and try 69 00:05:11,866 --> 00:05:16,866 to look at how -- do they tell you the same story? Are they converging? 70 00:05:19,100 --> 00:05:22,066 It's very confusing, I think, for many decision-makers to navigate through some of the models. 71 00:05:24,466 --> 00:05:26,500 MAN: We're going to start off with this weekend. 72 00:05:26,500 --> 00:05:31,266 MILES O'BRIEN: Weather forecasters are some of the most adept at navigating the inherent 73 00:05:31,266 --> 00:05:32,866 uncertainties of modeling. 74 00:05:32,866 --> 00:05:35,000 MAN: Going to have some travel problems if... 75 00:05:35,000 --> 00:05:39,333 MILES O'BRIEN: After all, it's been 70 years since they first ran a model through a computer 76 00:05:41,333 --> 00:05:44,500 to create a forecast. It's been steady improvement ever since. It's now possible to reliably 77 00:05:45,900 --> 00:05:50,433 forecast seven days in advance with 80 percent accuracy. 78 00:05:50,433 --> 00:05:55,433 But, with a novel virus, there are so many unknowns. And weather models do not have to 79 00:05:56,900 --> 00:06:00,466 account for human behavior. 80 00:06:00,466 --> 00:06:05,100 Marshall Shepherd is director of the Atmospheric Sciences Program at the University of Georgia. 81 00:06:05,100 --> 00:06:07,133 MARSHALL SHEPHERD, Atmospheric Sciences Program Director, University of Georgia: It's very 82 00:06:07,133 --> 00:06:11,700 important, when consuming these coronavirus models and weather models, to consume the 83 00:06:11,700 --> 00:06:13,733 uncertainty that we know is inherent. 84 00:06:13,733 --> 00:06:16,966 But we have a way to get around that in weather called ensemble modeling. 85 00:06:16,966 --> 00:06:21,866 MILES O'BRIEN: Ensemble modeling, meaning combining the predictions of many different 86 00:06:21,866 --> 00:06:26,866 models, it's a crucial tool that has greatly improved forecasting the weather and, in the 87 00:06:28,200 --> 00:06:32,233 past three years, seasonal influenza as well. 88 00:06:32,233 --> 00:06:37,233 Nick Reich is an associate professor of biostatistics at the University of Massachusetts-Amherst. 89 00:06:39,166 --> 00:06:43,500 Working with the Centers for Disease Control and Prevention, he leads a team that builds 90 00:06:43,500 --> 00:06:47,766 ensemble models to improve predictions of the spread of the flu. 91 00:06:47,766 --> 00:06:50,933 NICK REICH, University of Massachusetts-Amherst: I don't think any one model should be viewed 92 00:06:50,933 --> 00:06:53,000 as gospel truth. 93 00:06:53,000 --> 00:06:57,666 When you just use one model, you end up with a too strong reliance on one particular set 94 00:07:00,200 --> 00:07:04,166 of assumptions and one particular viewpoint. And this is why it's really critical to consider 95 00:07:06,266 --> 00:07:08,733 multiple models together. 96 00:07:08,733 --> 00:07:13,633 MILES O'BRIEN: The influenza models are informed by up to 20 years of experience with the viruses 97 00:07:15,300 --> 00:07:18,066 and the accuracy of the models. 98 00:07:18,066 --> 00:07:23,066 Reich and his team have now built a COVID-19 ensemble model. But it, of course, does not 99 00:07:24,233 --> 00:07:26,233 have the benefits of a long backstory. 100 00:07:26,233 --> 00:07:31,100 NICK REICH: We do have hundreds of years of theory about how to build mathematical models 101 00:07:33,433 --> 00:07:38,433 of infectious disease, but have they ever been tested in real time in this way, with 102 00:07:40,766 --> 00:07:44,166 all of the data sources that are available to us? No. 103 00:07:44,166 --> 00:07:48,833 We're building this car as it's careening down the highway, and we're learning about 104 00:07:48,833 --> 00:07:51,066 these models as we go. 105 00:07:51,066 --> 00:07:55,433 MILES O'BRIEN: Infectious disease modelers are scrambling to figure out where we are 106 00:07:55,433 --> 00:07:58,900 headed, depending on the decisions we make. 107 00:07:58,900 --> 00:08:03,900 If we take the time to better understand what the models can and cannot do, maybe we will 108 00:08:05,333 --> 00:08:09,500 do the same as we search for the path back to normalcy. 109 00:08:10,600 --> 00:08:12,166 For the "PBS NewsHour," I'm Miles O'Brien.