Portal de Periódicos da CAPES

A Conversation with … Kirk Goldsberry PhD, the Cartographer Whose Maps Revolutionized How Professional Basketball is Played

2020; Lippincott Williams & Wilkins; Volume: 478; Issue: 9 Linguagem: Inglês

10.1097/corr.0000000000001410

1528-1132

Seth S. Leopold,

Medical Coding and Health Information

Confession: I’m not a basketball fan. My apologies to those who love the game. You see, I’m from New York, the land of the Knicks. I trust I need to say no more. Fine, I’ll say more: How can a team like the Knicks, with the money and draw of a big-city market, and the heritage of an original NBA franchise (Red Holzman! Walt Frazier! Dave DeBusschere!), hire Hall-of-Fame coach Phil Jackson (with 13 championship rings) as the team’s president, and then go 80-166 in his three seasons at the helm? Now I’ll really say no more. Maybe just a little more: My hometown Knickerbockers have been just as bad even after they fired Jackson (67-163), bungling the offseason with amazing consistency every year since his departure. Even diehard Knicks fan Spike Lee quit going to Madison Square Garden. That’s it. No más. I can’t watch the stuff. If you grew up in New York and saw what I saw, you wouldn’t, either. But I do like baseball. I may not be a passionate fan, but I am, at least, a long-suffering one; I married into a Mets family. So, when Michael Lewis—a favorite non-fiction writer of mine—wrote Moneyball [5], I looked forward to reading it. To my surprise, although it was a “baseball book”, it wasn’t really about the game. Rather, Moneyball is a recipe for finding value where others miss it. The lessons of that book apply equally to orthopaedic practices and hospitals as they do to Major League Baseball teams. If your hospital awards surgical block time based on “percentage of minutes used” rather than on some measure of value returned—say, RVUs (or dollars) per hour—then buy a copy of Moneyball for your local hospital administrator’s next birthday, and read it aloud to him or her. Awarding surgeons for operating inefficiently makes as much sense as measuring the productivity of a car factory by sheet metal consumption rather than by the rate at which cars come off the distal end of the line. Back to basketball. It turns out there is a cartographer who has done with maps what sabermetrics (the math behind Moneyball) did for baseball, only in a much more comprehensible way (Fig. 1). And in the process, some fans of the court game say that this geographer, Kirk Goldsberry PhD, of The University of Texas–Austin (as well as Team USA Basketball and ESPN, and formerly of the San Antonio Spurs), broke the sport [7].Fig. 1: Dr. Goldsberry recently created a heatmap, which effectively shows Michael Jordan’s high scoring efficiency from inside the 3-point line during the 1996-1997 season. Published with permission from Kirk Goldsberry PhD.Dr. Goldsberry is the author of the New York Times bestseller, SprawlBall: A Visual Tour of the New Era of the NBA [3], which is a “geographic” history of the NBA’s 3-point line. His gorgeous heat maps (Fig. 2) use cartography to develop an evidence-based approach to shooting decisions. Among other things, Dr. Goldsberry found that players are about as likely to sink mid-range jumpers (about 15-18 feet from the basket; worth 2 points) as they are 3-pointers from the corner of the court (22 feet from the basket). The result? The NBA is in the middle of a 3-point boom. The 2017-2018 Houston Rockets became the first team in NBA history to shoot more 3-pointers than 2-pointers over an entire season [4]. The next season, they accomplished that same feat again—amazingly—shooting more 3-pointers than 2-pointers in more than 80% of the games they played [7]. This past season (2019-2020), Houston decided to do away with the traditional big man, trading their 6’11 paint-clogging center and replacing him with a 6’6’’ forward who can, you guessed it, shoot 3s. Today’s game is no longer plodding centers rambling towards the low block to setup their one-handed hook shot. Gone are the days of Kareem, Hakeem, and McHale. Today’s big man better move like Giannis (6’11”), shoot like Durant (6’10”), or defend like Embiid (7’2”). Adapt or die.Fig. 2: This heat map contrasts the most common shot locations from two different eras— the 1997-1998 season (left; labeled “[Michael] Jordan’s NBA”) and the 2019-2020 season (right; labeled “Today’s NBA”). These maps represent the evolution from the “Jordan era”, which generally prioritized getting close to the basket and taking mid-range jumpshots, to today’s “3-point era”, which puts a premium on taking more shots from beyond the 3-point line. Published with permission from Kirk Goldsberry PhD.He may have broken basketball, but I believe Dr. Goldsberry's approaches can fix health care. Graphic displays of quantitative information can enlighten [1] or grossly mislead [2], and even cause data to speak with eloquence [8] that when presented in tabular form would but lie there, mute. Please join me in the interview that follows with Kirk Goldsberry PhD, the cartographer whose maps revolutionized modern basketball, and whose approaches have the potential to do the same for healthcare. Seth S. Leopold MD:Set the scene. You’re a cartographer who’s taught geography at Michigan State. You’d done a tour as a visiting professor at Harvard. Who did you have to convince that you could use maps to change the way that basketball would be played, and how did you do it? Kirk Goldsberry PhD: Well, I went straight to the nerds. I presented my first basketball paper at the 2012 MIT Sloan Sports Analytics Conference in Boston, MA, USA. The paper argued that maps (spatial and visual analytics) have been vital tools in virtually every spatial strategy domain for centuries. Whether we’re talking about public health, military operations, or urban planning, maps are among the most important reasoning artifacts in world history. My paper asked why we weren’t leveraging them in pro basketball. Dr. Leopold:Your observation that basketball players were about as likely to sink 3-point attempts as 2-point shots revolutionized the way the game is played. The data were there all along; how did your cartographic displays change how those data were perceived? Dr. Goldsberry: We are visual creatures, and maps have a profound effect on learning. As you say, the data were there for decades, but something about seeing the numbers laid out atop a basketball court in cartographic form really made many people apprehend the concept for the first time. It made it more real. Dr. Leopold:In one commentary on your book, Sprawlball, the writer[6]suggested that Moneyball-style math—simple statistics on paper—suffice for some applications, but are utterly inadequate for more-complex relationships, which call for the kinds of visual approaches you’ve become known for. Where in health care do you think we can get away with playing Moneyball, and in what settings will the veil only be lifted when we start to use visual analyses and maps like yours? Dr. Goldsberry:Moneyball is a book about applying concepts and processes from the financial world to pro sports. It obviously struck a chord. I am passionate about unlocking new knowledge by applying visualization techniques to spatial data sets beyond the traditional subset of spatial scales they are normally applied to. Basketball is one such example, but I am 100% confident that medical imaging could use its own visualization revolution. Whether it helps scientists uncover new understandings or helps practitioners communicate more clearly with patients, visualization should be an emphasis in medicine. Dr. Leopold:What would it take to make that happen? I have to imagine that if I tried to create a Goldsberry-styled map it would work about as well as a cartographer’s first try at knee replacement. So how, specifically, can physicians and clinician-scientists turn your goal of emphasizing the visual in medical displays—especially the ones we use to educate patients and other physicians—into a new reality? Dr. Goldsberry: As someone who has benefited from knee surgery, I’m sure glad a cartographer wasn’t conducting the procedure. Still, just as there are grammars and expertises unique to surgeons, there are some unique to visualization as well. With that in mind, it would take collaboration between disciplines. As datasets grow larger and larger, it’s imperative that our interfaces to those datasets become more and more efficient. Visualization provides one great way to build windows into the kinds of large datasets that are defining the data-age of modern medicine. But all windows are not created equal. In my estimation, we need to foster interaction between mapmakers and medical experts to generate the kind of maps and interfaces that truly help us unlock the potential of the big data sets in your field. Dr. Leopold:Can you share with us some of the most-effective visual displays of healthcare-related data you’ve seen, and tell us why you believe they worked? Dr. Goldsberry: First, I think ultrasounds of fetuses are an example of how visual displays can transform minds. For many first-time parents, seeing those first images is life-changing. Similarly, I would also add that MRIs have not only changed the way doctors can diagnose and treat patients, but also changed the way they can communicate with their patients. Again, human beings were built to consume visual content and the more we exploit that basic fact, the better our reasoning processes will become. In both of those cases though, I still think your community has a major opportunity to improve what these images actually look like. The cartography is lacking! Lastly, from an epidemiological perspective, I think the COVID-19 curves in Italy and New York City will go down in history as somber yet iconic and lasting images of the Coronavirus outbreak. Dr. Leopold:Yes, I agree those curves were dramatic, but they weren’t really maps. Let’s talk about some of the maps that were used to describe the COVID pandemic; the good ones were interactive, some as stand-alone items (Fig. 3), while some other maps were integrated into more-complex dashboard-style displays (Fig. 4). Can you walk us through what works well in these two cartographic displays, and what you might do differently if you were making them?Fig. 3: An interactive map of the United States with a color-coded range of risk by county of hospital overload of patients with COVID-19. Published with permission from CovidActNow (https://covidactnow.org).Fig. 4: A screenshot is shown of John Hopkins University’s COVID-19 dashboard with heatmap and underlying data. This image was taken in mid-April. Published with permission from Johns Hopkins University (https://coronavirus.jhu.edu/map.html).Dr. Goldsberry: First of all, the pandemic has operated heterogeneously around the globe, and right away it became clear that maps would help us understand the landscape. The map in Figure 3 has an obvious takeaway—some American counties are experiencing very different predicaments than others, and it’s quite clear where the most-dire predicaments are. As usual, some maps have been very useful, including the ones we see in the New York Times, which does some remarkable interactive cartography. Similarly, the folks at Johns Hopkins deserve a lot of credit for building such a thorough dashboard of key information. Although the global map that you captured here is very coarse, their interface enables users to zoom in and interact with numerical data relatively easily, and that’s key. The biggest issue with both of these maps isn’t the cartography, it’s the reliability of the underlying data. There’s an old saying, “garbage-in, garbage-out,” and while people tend to trust maps more than they should, as you know, reliable data around this pandemic has been mighty hard to come by, and unreliable data simply presents as unreliable cartography. Dr. Leopold:My last question is for the basketball-junkies. Your shot maps are visually appealing. Watching teams shoot 3-pointers for 2 hours is not. In your book, you argue that nothing short of changing the shape of the basketball court itself can slow down the “3-point era”. Of the suggested court reconfigurations you proposed (thinning the lane from 16 feet wide to 12 feet wide, eliminating the corner three, and custom courts), which one would you like to see implemented and why? Dr. Goldsberry: I would love it if the NBA allowed teams to draw their own 3-point lines, making unique shooting situations in every arena. Shooters are too comfortable, and if every arena had its own 3-point situation, it would add some diversity to strategy, which is something we need more of in today’s game.