Data science has made key contributions to the fight against COVID-19, from tracking instances and deaths to understanding how populations translate travel restrictions into vaccine design. Harvard’s data science initiative aims to help university members, students, and fellows design and apply computer science and statistics equipment, and build a network to encourage the flow of ideas. The year-long Harvard Data Science Review published a special COVID-19 Engaging Factor online this summer that will be up-to-date with the latest findings, with the goal of fostering innovation and maintaining the verbal exchange. how knowledge science can help meet the challenge of COVID-19. The Gazette spoke with Francesca Dominici, Clarence James Gamble Professor of Biostatistics, Population and Data Science at Harvard’s TH Chan School of Public Health and co-director of the initiative, and Xiao-Li Meng, lead editor of the journal, and Whipple VN Jones Professor of Statistics at the Faculty of Arts and Sciences, on how the science of knowledge can be used to face the demanding current situations and, therefore, the demanding situations that the domain faces.
GAZETTE: What is the science of knowledge like for our understanding and reaction to COVID-19?
DOMINICI: Data science is probably on the cover of the New York Times every day. I believe that the pandemic has definitely increased the appreciation of knowledge science as a vital field that can help us solve high-demand situations that impact society. Data science is critical to perceiving nearly every critical facet of COVID-19. This includes proposing and testing a COVID-19 vaccine, perceiving the points that slow down the infection rate, perceiving the role of airborne transmission – which is key to perceiving if we can reopen schools – the identity of the environmental and socio-economic points and monitoring of mobility to better perceive the main behavioral interventions that involve the spread of the virus. Some of my studies, for example, focus on pollutants and COVID-19, which California’s wildfires are getting worse. It is difficult to think of an example related to COVID-19 that does not have a knowledge science method and demanding situations in the foreground, and Harvard University of all schools has conducted cutting-edge studies for the intersection of knowledge science and COVID-19. .
GAZETTE: How has knowledge science helped policymakers and others more obviously about uncertainty?
MENG: If there is a positive aspect of COVID-19, it is that it makes everyone aware of the importance of perceiving uncertainty. How do you rate uncertainty? How do you plan in uncertainty? For this special factor in COVID-19, we have introduced a new feature called “Executive Conversations”, and the first interview is with Preaspectnt Larry Bacow. We asked him how he uses the knowledge to plan for Harvard’s closure in March and how to reopen this fall. He says it was less difficult to close because the threat was asymmetrical: If we closed too early and it was nothing, he would have laughed at him, but the leaders laughed at him all the time. But if Harvard closed too late and other people died, we can’t live with that. Deciding how to open was much more complicated as there were many more unknowns. Students tend to be younger and less affected by COVID-19. But Preaspectnt Bacow had to worry about the entire university network: faculty, staff, and other age groups. It was incredibly difficult. We had a moment of verbal exchange with MIT candidate L. Rafael Reif, and we asked him the same question. MIT designed its dormitories to help academics interact with each other, which has now become a challenge. We have communicated how you can communicate with experts to perceive the threat, but due to the uncertainty, no one knows for sure. Collectively, we hope we can get a better picture, I don’t think we’ll ever get the best picture, and Harvard Data Science Review is a position to hear all those other voices and issues of view.
GAZETTE: Many others have been dealing with uncertainty, but the public may not fully perceive the central role that uncertainty has played in this pandemic. Leaders are forced to make decisions based on imperfect and even contradictory information. helping in conditions where there is no yes or no answer?
DOMINICI: We all feel the importance of quantifying and communicating uncertainty and accepting the desire to make decisions in uncertainty. Unfortunately, some leaders want to dismiss uncertainty in decision-making, while knowledge scientists want to recognize uncertainty, which does not mean that they [scientific knowledge] provide new data and make decisions by consultants. The result is a great tension.
GAZETTE: Is there a false impression that uncertainty means you have to reject the conclusions, because we are not sure, even if the uncertainty in your box only means that you have all the equipment at your disposal to locate a maximum probability route, perhaps the maximum likely path, to succeed?
MENG: We had this verbal exchange with the BBC news statistics officer on precisely this point. As knowledge scientists or statisticians, we like to provide things called “periods of trust. “We say, “We don’t know what it is, but there’s a fork. “But, ironically, providing periods of trust can cause the public to lose confidence in us. Many other people need a number, even though the truth is that we cannot produce a single number, because even the most productive imaginable number comes with a lot of uncertainty. We had a verbal exchange with Brief19 editor-in-chief Jeremy Faust, Harvard professor and emergency physician. He said it was incredibly difficult to estimate how many other people actually died from COVID19. You might think that this is a trivial question, but we know for sure at the beginning of the pandemic that other people have died whose deaths have not been attributed to COVID-19. Now, however, there is the option of over-attribution, because whenever other people die from multiple imaginable causes, if any of them are COVID-19, it will most likely be reported.
GAZETTE: I know this is an example of a broader point, but is there an exciting debate about COVID-19 death estimates Do you have any concept in which knowledge science is pushing numbers, above or below official estimates?
MENG: Well, to answer that as a true statistician, I don’t accept as true with any number of singles because they deserve to be given as a beach, another thing that makes this incredibly complicated is the quality of the data, an HDSR article cited through the World Health Organization, “On the identification and bias of bias in estimating the mortality rate of COVID-19 cases” , addresses several statistical bias resources when calculating the number of death cases. those other scenarios, and then see what the numerical diversity is. In a way, you can already see how the media has constantly reviewed the numbers. Although they report only one digit at a time, revisions reflect various types of state intents.
DOMINICI: There are two massive complications. First, it continues to evolve because we are still at the center of the epidemic, that knowledge continues to come, so all these analyses will have to be implemented to be repeated regularly, but I think the biggest challenge is that when you think about diversity, the number you decide on diversity has massive political and economic consequences. That’s why the role of knowledge science and the role of the Data Science Review is to be transparent about those challenges, so when looking back on the contribution of knowledge science to this, it is transparent that we have been rigorous and that we have not been partisan in one way or another.
GAZETTE: Have there been any key discoveries in knowledge science that have gained enough attention in recent months?
MENG: The main result, which many others in the room would suspect from the beginning but have not been emphasized enough, is that the quality of knowledge is very low. We all sense that no one is to blame because we are all suffering. and it’s hard to gather knowledge well when everyone’s looking to save lives. Any knowledge you can collect, you collect. In the medical community, there are practices to attend emergencies, we have emergency protocols, emergency rooms, etc. In the knowledge science community, we don’t have that concept of an immediate reaction team, so when something like this happens, we’re not prepared. We need percentages of knowledge without invading privacy, but how do we collect accurate and timely knowledge when other people are desperately looking to save lives?For top doctors, they don’t plan to gather knowledge, however, if you think about the total situation, gathering reliable knowledge also means saving lives.
Another question I think other people are starting to pay more attention to is how to deal with social dilemmas like privacy. Tracking other people’s movements is helping to perceive the evolution of the pandemic, yet here are huge privacy disorders. Do you locate the right balance? We used to have rules that we could simply paint with, but this pandemic is global and other countries have other tactics to do so. One article in particular, “Fighting COVID-19 Through Guilty AI Innovation: Five Steps in the Right Direction. “you’re getting a lot of attention. This is the longest article we’ve ever published, more than 16,000 words. The writer has established guiding principles to deal with these complex disorders, delicate disorders that Array does not have a solitary solution. These are charged disorders and, in after all, are not disorders that knowledge scientists – or any other organization – can solve. This is a question for society: what commitment should we have?
DOMINICI: Going back to Xiao-Li’s original point and that has not gained enough attention, there is no science of intelligent knowledge without intelligent knowledge. I think we are learning, but we want to do more to ensure that knowledge is available. A national registry of individual instances of COVID-19 will be made available. Some states publish knowledge and others do not. The vast majority of studies on COVID-19 have been done with the knowledge of the Johns Hopkins site. They have been at the forefront, however this knowledge is on the county seat of the United States and we would like to see individual knowledge. This comes down to what Xiao-Li pointed out in terms of mounting an emergency reaction to gather quality knowledge. There is no simple solution, but I think it is something we are working on. We also want a foreign registry of individual instances and deaths from COVID-19. There are privacy considerations related to knowledge of mobility, but there are fewer problems related to knowledge of cases because it can be anonymized. We want age, race and gender. Politicians make decisions based on evidence, which is why we want to obtain the most productive evidence possible.
GAZETTE: I was talking to other people about synthetic intelligence and COVID and they said the same thing. AI was more or less sad in our COVID response, and the explanation is that the quality of knowledge is very low.
DOMINICI: These algorithms are not wise if you don’t exercise them with high-quality data, you’re going to have synthetic synthetic intelligence stupidity.
MENG: The challenge is that the incentive structures are not adequate. Gathering knowledge well doesn’t make you a hero, but knowledge itself is essential. Not long ago, I had a verbal exchange with some other people who were deeply concerned with the production of national knowledge and statistics. I asked them what big reform they wanted to see, and their first reaction was the knowledge of the physical fitness record. Collecting this knowledge is not simple as there are other things involved, but still the knowledge itself. Many of us, unfortunately, suffer from multiple ailments, and doctors deserve to find out which one is the main one based on their medical judgment. In maximum cases, [it will probably happen]. But there are incentives to designate as condition number one that which is the maximum maximum likely to obtain the highest insurance reimbursement. It’s incredibly confusing, but most of the time we don’t hear about the complication, we just hear the results – how many instances have been reported. But analysis and forecasting is carried out without knowing what the underlying figures really mean. We want a national protocol to do those things. Another important factor is that you want a workforce that is well-trained enough to be at the forefront of gathering knowledge. They want to be able to read about the knowledge, know when it “doesn’t look good” and perceive that the decisions they make in collecting it will have a direct effect on the next analysis. Efforts are being made lately to provide education such as defined in “Data Shift: A Data Analysis Training Program for Public Servants. “
GAZETTE: Why not communicate about the origin of the Harvard Data Science Review?Francesca, why did the initiative arise that having a publication like this was a good idea?
DOMINICI: Harvard Data Science Review has been a conceptual medium for communicating the science of knowledge to the world. To take a step back, the Data Science Initiative was introduced in 2017, and its purpose is to paint in schools and departments to have interaction and activate the pioneers of knowledge science to tackle the main demanding situations faced by the humanities. create a highly collaborative network of studies to multiply the effect of the discovery of the science of knowledge in academia and in our society. The Data Science Initiative focuses primarily on studies and organizes educational conferences. We have a very successful corporate club program. We seek to unite our leading scientists, statisticians, and computer experts in the fields of law, business, public policy, education, medicine, and public health. So we were surely very happy when Xiao-Li came up with this concept to start a newspaper. It has become quite transparent that the science of knowledge is not just statistics; it’s not just you; it is actually a new field in which we have to integrate and take advantage of experience in other fields.
GAZETTE: Who is the target of criticism?Scientists?
MENG: Data science has this massive ecosystem, as I wrote in my first editorial. In the minds of many other people, the science of knowledge is device learning, computing, and statistics. But that comes with moral issues in gathering and analyzing knowledge, epidemiologists’ paintings on COVID-19, AI, and topics even quantum computing. Because other people who work with the science of knowledge are advancing in their fields of expertise, there is rarely a single position that combines to exchange concepts and effects related to the science of knowledge. Regarding the content of the Journal, surely we need clinical research, because it is vital that the science of knowledge is based on rigorous theories and methods. We must also highlight the impact, because the science of knowledge would not exist without its impact. And since we are a university, it is certainly essential to accompany the science of knowledge coaching. When a marketing team asked, “Who is your target audience?” and I said, “We’re targeting everyone,” they said I was crazy. But that’s literally what the science of knowledge deserves to be.
GAZETTE: Can you solve a typical problem?
MENG: The magazine has 4 main sections. “Panorama” features articles by thought leaders on everything similar to the science of knowledge: philosophy, industry, government. Cornucopia presents impact, innovation and transfer of wisdom, highlighting how the science of knowledge can be used in any field. Then “Stepping Stones” includes learning, training and communication. The last one is “Milestones and Millstones”, where the deepest curtains sink. We also have columns with other topics. An actress from the UK has written an existing version, in which she explains how the statistics ‘stop showing those curves’. There are columns aimed at pre-college students, aimed at the general public, like “Can Machine Learning Predict the Price of Art at Auction?” and “Recipes for Success: Data Science in the Home Kitchen. ” We have columns on the history of AI, the history of baseball. The purpose here is that anyone can spot that problem, any problem, and locate at least one article where they say, “Well that’s interesting. ” You can read articles that don’t have a formula, then go ahead and think, “My gosh, how can anyone read this?” It is essentially like a magazine published in several languages. What you get depends on who you are.
GAZETTE: Where does the initiative happen during the year?
DOMINICI: We have replaced the course of this year by what is happening with COVID-19 and what is happening with racial discrimination. These are things we need to pay attention to. I was inspired when we were contacted through our Harvard Data Science Postdoctoral Fellows, who said, “We really need to think about the role of knowledge science in tackling racial bias. ” Therefore, the goal of the initiative is to pay even more attention to these broader concepts through the prism of the science of knowledge. We have announced a series of activities spanning the science of guilty knowledge and the science of knowledge showing a discrimination bias. We are dedicating a series of seminars and investment studies to using the science of knowledge to uncover biases and to perceive and address the use of poorly designed knowledge science that reinforces bias and inequality. There are many examples where if you exercise the device learning models used in synthetic intelligence, for example in genetic or diagnostic knowledge of the white population, you cannot draw conclusions about what is happening to the population. black. We all know examples of corrupt justice that have exacerbated the stigma. We also have a very strong corporate club program and some other flagship initiative on Trust in Science: How can public acceptance as true in science be fostered by harnessing the science of knowledge? For example, what position will other people be in to take the new COVID vaccine?
This interview edited for clarity and extension.