麻豆学生精品版

Skip to main content
Mole Crowdsourcing to Detect Skin Cancer

You are listening to Health Library:

Mole Crowdsourcing to Detect Skin Cancer

Apr 15, 2016

Imagine this: during a skin self-exam you notice a growth or mole that looks suspicions but you aren鈥檛 sure. You take a picture of it with your phone using a special app that allows others to vote whether or not they think you should do something about it. The next day, you wake up and learn that 38 percent of the people that looked at it you should see a dermatologist. Sound crazy? Maybe, but it works. is a communication researcher at University of Utah who won a $2.2 million National Institutes of Health Director鈥檚 New Innovator Grant to implement a system just like that. He talks about how he came up with the idea and why it works.

Episode Transcript

Interviewer: Mole crowdsourcing: it's an innovative way to discover and identify melanoma and that's coming up next on the scope.

Announcer: Examining the latest research and telling you about the latest breakthroughs, The Science and Research Show is on The Scope.

Interviewer: You know, it sounds a little gross. Take a picture of what you suspect to be a cancerous mole and then post it to the Internet for others to see, but it turns out that might be a really effective way to screen for melanoma. Jake Jensen is from the Department of Communication at University of Utah and he won a $2.2 million grant, part of the 2015 and NIH New Innovator Grant to look at that. So first of all, is that an accurate way to screen for skin cancers, take a picture and have non-trained people look at it?

Jake: Well, a dermatologist might quibble with the phrasing of that. Whether it's a form of screening or not, we could debate. What it is is it's a really effective way to potentially move people to go to a dermatologist.

Interviewer: So let's talk about the way things are and why that's not necessarily getting the job done and the way that you hope your research points to the way things could be.

Jake: What happens right now? Well, right now, we say to people, "You should engage in monthly skin self-exams where you look at your body and look for strange moles, strange growths. And when you see them, you should make note of them and you should go into a dermatologist for a clinical skin examination." That's our basic game plan.

There are lots of problems with that game plan. The first problem with a game plan is skin self-exam is not very effective. People are very bad at finding odd-looking lesions and moles on their body. And so I spent the better part of a decade doing research on skin self-examination, trying to improve the technique. And study after study, people were horrible at it. No matter what we did, they were horrible. The only consistency in the research was no matter what we did, people were horrible at skin self-examination and that's what other researchers were finding as well.

We were all frustrated in one day, in my frustration, I walked into one of my colleague's offices because he has a wonderful couch in his office. And I flop myself down on his couch and I said, "I'm so tired of skin self-examination research. No matter what I do, it won't work. There is no solution." And we talked about it for a while as I vented and eventually, he said something that forever changed the way I thought about this. He said . . . keep in mind, he's not a health researcher. He's somebody who studies new media so he was a good person to vent to in that he was like, "I don't know what you're talking about, but I know you're angry." And he said, "Well, is it that individuals are bad at skin self-exam or is it that groups are bad?" And I thought, "Individuals are bad so groups are bad. So I don't understand the question."

Interviewer: Yeah. Because if one person's bad, how are 20 of those people any better?

Jake: I said, "I don't understand the question." He goes, "Well, there's this thing called collective effort that says sometimes a group is good at something when an individual is not. You can take a group of individuals who are bad at a task, but as a group, the group somehow can be used and mobilized to be good at a task." And I said, "Okay. Rather than, 'Are individuals good,' here's what I want to ask." I took a rock or a curve. It's a type of statistical analysis. I said, "Is there a rock curve that fits to this data?"

And there was, at the group level. And here's what it looks like, for most moles that people look at, they're not suspicious. Take a photo of any mole on your body or any mole that you encounter. Take any mole imagery and you show it to people, generally, they say, "That looks fine." However, when more than 19% of people say a mole looks suspicious, now we're in a different world. If you use that as a cutoff, you can find 90% of melanomas because there's something there that the group picks up on. Here's why I kept missing it: because when you're thinking about the individual level of ability, you're saying, "Well, I want individuals to find it 90% of the time." But individuals can't do that. Groups can when we use a cutoff and we say, "Well, if it's more than 19% of the group that's concerned, let's use that as an indicator."

Huge implications for that. It doesn't matter whether you train people. In a sense, the best group is a group where you say, "Just tell us whether you think that's a weird image or not. Is that a weird mole? Just yes or no." Sometimes, people would say to me, "Can we have laypeople doing this?" It's risk factor. They're not trained. When it comes to laypeople, I see no evidence that training them makes them better at this intuitive, sort of "Is it weird or not" because I'm only going to look at them on the group level. I'm not going to look at them as individuals. So I just want to know if the herd finds it weird. And the answer is when they do, we kind of find melanoma.

Interviewer: Let's go back to the original problem. The original problem is individuals aren't good at self-screening. So they see a mole and maybe they might not think it's a problem. Is that the problem? Or is the problem that they don't do anything about it?

Jake: Yeah. Well, chicken or the egg in some ways.

Interviewer: I guess what I'm asking is will this solve the problem that you've laid out?

Jake: So here's what we want to do. I want to tell a real quick story because it'll help you to understand where we want to go with this. Imagine there's a farmer and the famer's name is Joe. Joe is out in the field and Joe sees that he has a weird mole on his arm. He comes home that night and he tells his wife, Martha, "Hey, I've got this weird growth on my arm." Martha says, "You need to go in to a dermatologist."

Joe is reluctant to go to the doctor for any reason and Martha knows this. So she rails on him for the rest of the night. But Joe, "Eh, it's harvest." He'll go later. Eighteen months pass by before Joe eventually goes in. By the time he goes in, he has late-stage melanoma. There's very little they can do for it at that point in time. Joe passes away within the next few months.

I'll give you an alternative ending to that story. Everything is the same about the story, except this: when Joe comes in from the field and shows Martha the mole, Martha pulls out her phone. She snaps a photo of the mole. She doesn't even bother arguing with Joe because she knows Joe will not listen. The next morning at breakfast, Martha slides the phone across the table and says, "Ten thousand people have looked at your mole and 47% of them think it looks weird." And that's such a high score that it was flagged in the tele-dermatology system and a dermatologist who does rural rounds, who will be within 40 minutes of us next week, wants to see you at 9:00AM next Tuesday because she's looked at the mole and she is suspicious as well."

Joe goes in next Tuesday. It's pre-cancerous. The dermatologist chops it off. Joe lives. Now, that cell phone system I just described is what we're trying to build.

Announcer: Discover how the research of today will affect you tomorrow. The Science and Research Show is on The Scope.