LC: Do you think that data science and computer science have an image issue in popular culture—that they’re not portrayed as passionate and engaging endeavors?
EL: For sure, for sure.
How does one tackle that? The army has TV and radio ads all the time with people jumping out of planes saying “We need people who can do this.” How can the computational sciences compete with that, short of an overly-dramatic ad campaign?
That’s a great question. Maria Klawe, who is now the president of Harvey Mudd, has for years been trying to get some TV network, TV probably doesn’t matter anymore, to do a program that shows the practice of computer science as attractive like LA law did being a lawyer or various doctor shows did being a doctor. All we’ve got is Silicon Valley which makes it look like hell.
I think part of our pitch for close to ten years now has been to think about the societal challenges that computer science is essential to tackling. The way I think about this is that we’ve spent fifty years making things smaller and faster and less expensive, but these days what people care about is healthcare and education and transportation and energy independence and scientific discovery and a set of things like that. There are more modern areas of computer science, whether machine learning and data science or visualization or robotics or whatever, that couple continued advances in the core of these societal challenges. Our pitch is that if what you care about is improving lives in the developing world, then the best way to do that is by utilizing and advancing computer science. If what you care about is the efficacy and cost effectiveness of healthcare, then you better care about computer science. If you care about education then you better care about computer science.
We have people in this building doing all of those things. A guy up the hall has spent years doing incredible work on personalized education, now being pushed out into all sorts of schools. He began by essentially trying to produce adaptive technology for teaching algebra to elementary school kids. What it does is not replace teachers, rather it’s personalized in that it figures out the state of knowledge and the conceptual blockers for each individual student and lets the teacher understand these and lets the teacher spend his or her time in a more effective way with each student, and the results are just amazing.
We have a guy down the hall this direction who has been turning smartphones into medical, diagnostic, and screening devices, so it does bilirubin, osteoporosis, and blood pressure from a smartphone—mind-blowing stuff. The blood pressure is an example of what you just wouldn’t imagine it’s possible. A year ago we had John Markoff from the New York Times here with a blood pressure cuff on one arm and a smartphone on the other, and they read the same thing. Here’s how the blood pressure works for example. It turns out there’s a relationship between the pressure in a fluid system and the rate of a fluid pulse flow through that system, that’s not surprising. It turns out that also the ratio of the distance from your heart to your temple and your heart to your fingertip is about the same independent of body size. You hold the smartphone front camera at your temple with your finger over the other camera, the camera can see the blood pulse, and you can determine the timing difference. Then the equation and the timing difference lets you use this pulse-flow-time-pressure equation to tell you what your blood pressure is, it’s mind blowing.
Bilirubin is tied to infantile jaundice, and it causes brain damage if untreated. It’s not so hard to detect in a Caucasian kid, although parents worry about it, but in dark complexion kids in the developing world it’s very hard to detect. It turns out that bilirubin in your blood absorbs a particular wavelength of light in the blue spectrum. So what you do is you use the light from the camera and the image sensor on the camera and, with some calibration that looks like a paint swatch of standard colors, you determine the depth of the notch in the reflected light at the point of the spectrum that bilirubin absorbs and you can determine the percentage of bilirubin in the blood—unbelievably cool stuff, so a bunch of this is going through FDA approval now.
It’s sort of like what we’re doing. I’ve being doing the scientific discovery part of it. The pitch is that you don’t have to care about making things smaller and faster and less expensive to want to do computer science. If you care about tackling any societal challenge, then computer science is where you ought to be.