Episode 1: WHOA, THAT WAS CLOSE! Can we prevent crashes from happening by studying the ones that almost happened?

Bernie Fette (Host) (00:16):
Hello! Thanks for joining us for Thinking Transportation, a podcast looking at how we get ourselves and the stuff we need from one place to another. I’m your host, Bernie Fette, editor at large at the Texas A&M Transportation Institute.

Bernie Fette (Host) (00:32):
Nearly 45,000 Americans die every year in roadway crashes. A hundred times that many are badly injured. Even with a pandemic lockdown, roads are just as dangerous. Cars are getting safer and traffic laws are getting stronger. And yet the average death count remains at about a hundred every day. If we intend to bring that number down, we might need a fresh approach. So it’s good that Shawn Turner and Eva Shipp are working on one. Eva is an epidemiologist at TTI and Shawn is a senior research engineer. Shawn and Eva, glad you could share your time today. What are you working on to make our roadways safer?

Shawn Turner (00:33):
We’re basically trying to develop ways that we can predict car crashes before the drivers actually crash their cars. Right? And so we’re doing that with what, what I’ll call near misses or, like, leading indicators. We’ll talk a little bit later about leading indicators from these newer cars that are connected to the internet, you know, cause it’s, it’s funny these days that some of the newer model cars actually know more about us as drivers than we’re willing to admit about ourselves.

Shawn Turner (01:50):
Um, you know, that’s like they know when we drive aggressively, they know when we change lanes without putting on our turn signal. Um, they know when we jam on our brakes, and the cars are talking to the cloud, at least the newer model cars are. Um, and they’re sending data about our driving using cell modems that are now included on, on almost all the new cars, like even the, even some of the modest economy model cars. Imagine that I’m driving my 2018 Ford F-150, you know, like all, all Texans have their Ford F150. Um, and I’m driving out between my different oil well properties, right? Because that’s what all Texans do. Um, and a large 18-wheeler pulls out right in front of me at this new oil well site. Right. And I’m jamming on my brakes because I don’t want to run into the back of this big oil tanker that just pulled out in front of me.

Shawn Turner (02:43):
Well, guess what? I just jammed on the brakes; that just got sent to the cloud. And when I say cloud, I mean the internet, this huge database that’s, that’s connected to the internet. Right? And so now imagine that there’s thousands and of other cars out there that are talking to the cloud whenever their drivers also jam on their brakes and maybe it’s drivers that are jamming on their brakes at this exact same oil. Well, I mean, like within a few weeks, it’s going to be pretty clear that this new oil well is generating a lot of these big oil tankers, right? And some of them are pulling right out in front of other Ford F-150 or Chevy Silverados or whatever. And so we accumulate this data over thousands of different cars. And we, at some point, we’re going to get a sense of where are people really jamming on the brakes and where is it more likely that crashes are going to occur hopefully before they occur. This is a real-world example of stuff that we’re doing right now in West Texas, where the energy sector, it’s really booming and traffic safety is, is a big problem. Even in this rural area.

Bernie Fette (Host) (03:57):
Well, interesting that you brought up that example because just a couple of days ago, I’m following a couple of vehicles. Both of them, you know, just plain old SUVs that we see every day. The leading car, we’ll say Driver One, had to jump on the brakes driver. Two, following a little too close behind, had to jump on his or her brakes almost as fast. Driver Two was not going to miss the back of that car in front of them. And so Driver Two swerves really hard to the right, just barely missing Driver One. So what happened in that case? What were the signals that were sent and how quickly were they sent to wherever they went?

Shawn Turner (04:42):
So in that particular example, most likely what happened is within a couple of seconds of when those drivers jammed on their brakes, if it was a newer model car, that car was probably talking to the internet and saying, “Hey, my driver just jammed on their brakes.” So, right now cars are mostly just talking to the cloud or they’re talking to the internet when something out of the ordinary happens. It’s not too far away, it’s in the near future, that cars are actually going to be talking to each other. Right. And so like in this specific example, Driver One slams on the brakes. Driver Two noticed that Driver One slammed on the brake ; well, three, four years from now the car, Car Number Two, is going to get a signal that Car Number One slammed on their brakes and Car Number Two may automatically apply the brakes without the driver, even kind of being tuned in. And that’s, I guess that’s the, that’s the scary thing is that as cars start to sort of drive themselves or they take more and more autonomy, there’s a concern that drivers are going to start to unplug.

Bernie Fette (Host) (05:55):
So this newer method that you’re describing gets the information to the people who make decisions about safety improvements faster than I guess it does currently. How, how are we making decisions now about where safety improvements are made? You know, a turn lane gets added or a speed limit changes. How’s that being done now?

Eva Shipp (06:16):
So, by and large safety improvements, now, and in the past have been currently based on studies of crash statistics or historical crash data or crash records. And that’s great for telling us all sorts of information about where crashes are happening. Who’s getting involved in crashes like different age groups and what-not, but some issues with that are some things that may impede rapid progress are things that would oftentimes we need large amounts of historical data to do a robust crash analysis. And that requires us a lot of time to accrue those events. And also there can be a lag in the amount of time it takes from the time that you collect that information to the time that you can actually analyze it. So, even though we’ve developed a lot of robust methods and been very successful at implementing safety interventions or safety improvements, using those historical crash records, it still requires oftentimes a considerable amount of time before you’re able to act, you know, really look at the information.

Shawn Turner (07:26):
Yeah. And so this is the interesting thing in our safety analysis is that this looking at crash statistics and looking at, at when and where people have had crashes, that’s, that’s been the tradition. That’s how we’ve done safety analysis for decades, but then you think about it and you kind of think, well, that’s kind of like, “Hey, let’s wait till we have a full on recession until we change the interest rates.” Right. I mean, so just like we have leading indicators in the economy, it just, it seems like we should have leading indicators in traffic safety, right. Something that gives us a little bit of a head’s up that something’s not quite right here. We can, we can kind of wait around for proof, but if we wait around for proof, we might hit the bottom, so to speak.

Bernie Fette (Host) (08:11):
Okay. That sounds really wise and practical, but it also sounds like at least some people might say, it’s a bit on the revolutionary side. If you’ve been doing things one way for decades, and now you’re suddenly suggesting something that’s not looking at history, but looking at what you might be able to predict. Are there skeptics out there?

Shawn Turner (08:20):
Yeah. Oh yeah. Most definitely. I mean, and so that’s, there’s a couple things I want to mention. One is, so I’m an engineer and I don’t know much about psychology, but I understand institutional inertia; that basically the resistance to change and the resistance to something different. And so, you know, I think traffic engineers are used to doing crash analyses. They’re used to looking at crash statistics. And with very limited money for road improvements, they typically won’t make changes until enough crashes accumulate. So that’s, I mean, that’s the first thing we’ve got to overcome is, can we convince traffic engineers that a problem exists before it becomes too big of a problem, before too many crashes occur?

Shawn Turner (09:18):
Um, I mean, there’s one or two others, there’s, you know, we’ve got to have the right leading indicators for a safety problem. And so, like, one of the examples that we’ve talked about already is people jamming on their brakes or what we call hard braking. Right. So hard braking may not be a very good predictor of crashes where people are simply, you know, like running off the road because they’re sleepy or they’re distracted. In cases like this, probably the leading indicator is going to be more likely the, sort of the side-to-side movement, you know, if someone is weaving or if they’re drifting over a lane line, but then certainly like in urban conditions, stop-and-go traffic hard braking–yeah, that’s probably a good leading indicator.

Eva Shipp (10:01):
You know, when I think about this issue from a public health standpoint, it kind of reminds me of, you know, many years ago, we didn’t really understand that smoking causes lung cancer. There were people that were getting lung cancer, but there was not a lot of recognition of why. And it took some very rigorous analyses to figure out that smoking was, in fact, contributing to lung cancer and that there were carcinogens in cigarettes. And that was pretty revolutionary thinking many years ago. I mean, people, even doctors, uh, opera singers would have big advertisement magazines saying how wonderful smoking is for their health, right? So it took a lot of very deliberate thinking and analytical work to figure out that smoking was a risk factor for this really costly health problem, just like injuries and crashes are a huge public health problem.

Bernie Fette (Host) (11:03):
I’m really glad you brought up that example because you’re an epidemiologist, Eva. And I think that before the pandemic that we’re in the middle of, an awful lot of people never even heard the word epidemiologist, and even fewer knew what such a person did. Shawn was talking a minute ago about some of the naysayers, perhaps in the traffic engineering community. Do you have naysayers too in your line?

Eva Shipp (11:31):
Most definitely. Um, you know, we’re talking about looking at near misses to help us prevent crashes. And there are definitely people who think that understanding near misses can be very helpful for injury and crash prevention. And then there’s others who maybe think, well, just because there was a near miss, that doesn’t necessarily mean that we were actually, if we, if we act on what we know about that near miss, would we necessarily be able to prevent an injury event. And so I think that’s why doing very careful, robust analysis and showing people what we’re talking about with data.

Bernie Fette (Host) (12:09):
So the one thing that both of you have in common is that convincing the naysayers is going to depend on the evidence that you can pile up. Right?

Shawn Turner (12:16):
Yeah, that’s exactly right.

Eva Shipp (12:17):
Absolutely.

Bernie Fette (Host) (12:19):
Okay. In addition to the personal tragedy of car crashes, they also carry a pretty huge economic cost. And so where does the economics of this idea that you’re working on play out? Is there a chance for big cost savings as well as life savings?

Eva Shipp (12:36):
Absolutely. In our work in my group, we looked at the National Safety Council’s estimates of economic costs a lot. And they look at, you know, what’s the cost of different injuries due to crashes–

Bernie Fette (Host) (12:51):
And, and what, I’m sorry, what, what is that cost instantly? I’ve seen it before. It’s pretty high, right?

Eva Shipp (12:56):
The average economic cost for every death, not every fatal crash, but every death is close to $1.7 million based on the estimates from 2018. And in those estimates, they look at everything from wage and productivity loss, the medical expenses, administrative expenses, motor vehicle damage, employers’ uninsured costs… So, you know, when we think about the impact of an injury from a crash, we think about automatically the medical expenses, but the impact is much broader on all these estimates that I just talked about… But the impact on one’s family? That’s not very easy to measure. And so, it’s a tremendous amount of savings that we could do by preventing these crashes.

Shawn Turner (13:46):
Yeah. And I I’ll add something that, you know, I think the value of this somewhat pales in comparison to the impacts on the driver, in the family because of the injuries. But I think having leading indicators and having more data to better diagnose safety problems. That’s going to help transportation agencies, right, to more quickly diagnose where the true problems are and what needs to be fixed. So, I mean, I, I kind of liken this to, you know, like taking, taking a car to a mechanic 40 years ago. Right. You take it to a mechanic 40 years ago. You got to describe “here, it makes this noise when it does this.” And then the mechanic’s got to spend hours and hours trying to sort of recreate that noise maybe that you heard, or that issue that you have, what do we do now? I take it to the mechanic. He plugs it, he plugs it into his computer … five minutes. He says, all right, Shawn, you’ve got loose brake pads and you got this and you got that, and I’m fixed in and out the door. So I, that’s kind the way I see at least another way that we’re going to be saving money with this approach.

Bernie Fette (Host) (14:59):
I wanted to come back briefly to the idea of even with data when you’re using that data to come to those predictive decisions, how do you take into account something so unpredictable as driver behavior?

Shawn Turner (15:14):
Yeah, that’s, that’s a good question. So I will tell you, this is why data science is one of the hottest new professions, and that’s why data scientists are so in demand, right? Is because data science allows us to see trends and patterns and things that, you know, to our human eyes and do our brain–it kind of just seems random and unpredictable. An everyday example, I go to Google search and I start typing in a few letters, and within a few letters, Google search is predicting what I’m searching for even before I type the whole search term. So it —

Bernie Fette (Host) (15:50):
That’s a little scary.

Shawn Turner (15:51):
All right. So, how can Google search predict what I’m searching for? You know, I’m a, I’m a random, unpredictable person, I could be searching for just about anything. Well, it turns out that Google collects data about me and you and others. And they use that data in very clever ways to predict fairly accurately within a few key strokes, what I’m going to be typing in the next few seconds. And then it’s kind of the same concept with car crashes. So connected cars gives us a chance to study and find patterns in driving behavior that may seem totally unpredictable.

Bernie Fette (Host) (15:56):
Well, and maybe with things like your cell phone in your car, maybe we as drivers aren’t as unpredictable as I was suggesting earlier.

Shawn Turner (16:37):
That’s right.

Bernie Fette (Host) (16:37):
Let’s say that later this year, you’re going to your college reunion, and a classmate that you haven’t seen for a long time asks you what you’re working on. And they were not in the same science field that you were. How do you help them understand what you’re working on? You’ve got 30 seconds.

Eva Shipp (16:58):
So, I think in a nutshell, I’d tell him, you know, I’m working on accelerating the rate at which we turn data into information and ultimately actions that save lives on Texas roadways.

Bernie Fette (Host) (17:10):
Well–

Shawn Turner (17:10):
That’s pretty awesome.

Bernie Fette (Host) (17:11):
Yeah. Lots of seconds to spare. Okay.

Shawn Turner (17:14):
Can I have some of those Bernie?

Bernie Fette (Host) (17:15):
You have to ask like they do in Congress, if the gentle lady yields her remaining time to you.

Shawn Turner (17:23):
So, I know my classmate isn’t a scientist, but I do remember that this particular classmate is a movie buff. So I say, “Hey Jane, do you remember that Tom Cruise movie that was called Minority Report?” Awesome movie. But in the movie, Tom Cruise predicted crimes before they were going to happen in the future. And then he stopped the crime before it was actually committed and people were killed. And that’s exactly what we’re trying to do with road safety. We’re trying to predict the crashes and injuries and fatalities before they even happen so that we can prevent tragedy from happening.

Bernie Fette (Host) (17:59):
Very good. Of course, you probably wouldn’t have done quite as well without Eva’s 17 seconds that she gave you. Why are you doing this? What’s what motivates you everyday to get up and work on this?

Shawn Turner (18:15):
I love data. I don’t know what it is, but throughout my 30-some-year career, I’ve always enjoyed working with data and getting more insight from data, learning new things. And I’ve done this in other areas where we’ve gotten new data sources and we’ve made quantum leaps forward. You know, I wouldn’t have been able to imagine, for example, how well we can measure traffic congestion now from cell phones. I just couldn’t have imagined that when I started my career, I want to do that in safety.

Bernie Fette (Host) (18:50):
What’s driving you, Eva?

Eva Shipp (18:52):
I think for me, too, it’s similar to what Shawn was saying,; but also these events, these crashes are preventable. And from a public health perspective, my whole career is spent on improving people’s lives, through improving their health, especially when these are things that we can keep from happening. If we understand better what’s making crashes happen. And if we can accelerate the rate at which we put in countermeasures and interventions,

Bernie Fette (Host) (19:20):
You both represent fairly distinct disciplines: engineering and epidemiology. I’m guessing that whenever you were both studying in college, you may not have considered that you’d be working together. So how do your disciplines complement one another and help us get to safer conditions on the roadways?

Shawn Turner (19:39):
Yeah, so to me, that’s one of the great things about TTI is that TTI researchers have the opportunity to collaborate and problem-solve with people that sort of view problems and solutions through a different lens than how I, an engineer, would view the world. So, I mean, how do you combine that expertise? I think one of the things that I’ve learned is that you have to respect everyone’s strengths on your team. So, just because they look at problems differently than you do that doesn’t mean that one approach is right and the other not right. I mean, I tend to look for some common middle ground. What perspectives can we bring from epidemiology, from engineering to get a more complete picture of possible solutions? And I mean, honestly when I’m on research teams and not just on this effort, I, in other topic areas, I ask myself, you know, what kind of expertise do we need to help make progress? I mean, even in this area, we need data scientists who are experts at making sense of these huge data sets. We need geographers who have these computer tools for analyzing location-based data. Uh, we need computer scientists there. We need experts who with black high performance computing, just to be able to manage the sheer size of our data set. So it really, I think these days, our world is so complex that we really have to take a multidisciplinary approach to make progress.

Eva Shipp (21:08):
And, you know, listening to Shawn talk. It reminded me of why I came to TTI. So, until five years ago or so I was faculty at a school of public health. And one of the reasons I came to TTI and made a pretty big transition was to be able to work with people like Shawn and engineers and folks that are from the discipline that I know very little about. I didn’t really understand what a roadway network was. I could tell you a lot about the demographics of people that get in crashes and the injuries that they sustain, but that was, you know, truly addressing this issue requires understanding and input from the engineering side. So that’s one of the real big joys of working at TTI, is it’s so easy to be in these multidisciplinary teams.

Bernie Fette (Host) (22:02):
It sounds like you really don’t have any choice that the problems just aren’t as simple as perhaps they used to be. And so the problem-solving can’t be either.

Shawn Turner (22:11):
Absolutely, amen to that.

Eva Shipp (22:15):
Yeah.

Bernie Fette (Host) (22:15):
Eva Shipp is an epidemiologist, and Shawn Turner is a senior research engineer and they both do research to make the roads safer for you and your family.

Bernie Fette (Host) (22:27):
More than 150 years ago, there were skeptics who said “you can’t forecast the weather,” but what started out as a revolutionary idea is now commonplace. And we use those predictions to save lives when hurricanes threaten. Can we do the same thing when it comes to car crashes? That’s another one of those revolutionary ideas, and we’re not there yet. But before long, we might be able to prevent crashes from happening by studying the ones that almost happened.

Bernie Fette (Host) (22:58):
Thank you for listening. We hope that you’ll subscribe to our podcast and share it with your friends and colleagues. We also hope you’ll check in with us again next time, when we visit with David Schrank, David’s an expert on traffic congestion, what causes it and what can fix it.

Bernie Fette (Host) (23:17):
Thinking Transportation is a production of the Texas A&M Transportation Institute, member of The Texas A&M University System. The show is produced and edited by Chris Pourteau. I’m your writer and host, Bernie Fette. Thanks again for joining us.