July 25, 2012 | 1
This year’s Google Science Fair winner, Brittany Wenger, 17, from Sarasota, Florida, spent more than 600 hours coding a sophisticated computer program to help doctors detect breast cancer using a less invasive form of biopsy. I spoke to her this morning at Google headquarters in Mountain View, California.
How did you feel when you heard you had won not only your age category but also the grand prize at Monday night’s awards ceremony?
I was just so excited. It was a very surreal experience walking up there. I don’t even know how I got up there.
Tell me about your project.
I taught the computer how to diagnose breast cancer so it could determine whether a breast mass is malignant or benign. I did this because currently the least invasive form of biopsy is actually the least conclusive. So a lot of doctors can’t use it.
I created an artificial neural network, which is a type of program that learns based on its experiences and mistakes, so it classifies problems that are far too complex for humans to classify. Then I fed information into the neural network from a database of fine needle aspirates, the least invasive form of biopsy.
Currently the network is working really well. It is 99.1 percent sensitive to malignancies, and I ran 7.6 million trials and proved that, as I get more data, the success rate increases and the inconclusivity rate decreases, so I think with more data it will prove to be hospital ready.
What inspired your project?
I started in the 7th grade. In school we were researching the future, and my part of the future that I was researching was future technologies. I grew fascinated by artificial intelligence, which I came across. I went home that night, and I bought a computer programming book and, with no experience, decided that was what I was going to do with the rest of my life.
Computer science is one area where the gender gap has not yet closed: men still outnumber women by a wide margin. Why do you think that’s the case?
I think sometimes there’s a stereotype around computer science, that it’s just video game development, and more boys are hard core game developers than girls. But you have to realize it’s our Web sites, our Google tools, it’s our Facebook, and I think that you could reach girls more if you could appeal to what they’re using computer science for.
But also I think we’ve come a long way. More girls are getting interested in science, and I know it used to be that girls weren’t encouraged, but I’ve never felt like I couldn’t go into science, like I was being discriminated against because I was a girl.
Do you remember what first sparked your interest in science more generally?
I always say that science is a passion that found me, instead of a passion that I chose. When I was little my brother was really sick with a lung condition he later outgrew, and we were in the hospital at all hours during his first few years. I really grew to idolize the men and women in scrubs and got a taste of what science could do to change the world. But along the way I had a 6th grade science teacher who was just amazing, her passion was so contagious, her name was Karen Malesky, and I’ve been blessed with amazing science experiences. It’s always been hands-on, and I really like that aspect.
Have you decided what career path you’d like to pursue?
I want to be on the frontier of cancer research, finding the cures that are going to save lives and doing things with computer science that can be the technologies of the future. I also want to be a pediatric oncologist, so I hope to intertwine my passions for research, computer science and patient care in the future.
Many of my readers are parents with science-interested kids. Do you have any advice for them on how to help foster their children’s interest?
Science is so broad, so let your kid follow their passion. My first artificial neural network had to do with soccer. I’m an avid soccer player. I taught the computer when it should pass, when it should dribble; it was a mini soccer game, two on two. When you’re passionate about something you’ll be persistent and really enjoy it.
What are the next steps for your project?
It will take a long time, but I hope to scale it up and bring it into hospitals. I put my neural network into the cloud, because the cloud is this amazing, elastic entity that allows for a million hospitals to access it tomorrow if they want and to provide feedback. I’m so happy to have won the Google Science Fair, because it will give me a new platform to help bring my project to physicians, and people will take me more seriously.
In March 2013, Brittany finished 8th and won $20,000 in the annual Intel Science Talent Search. I asked her what it was like to take top honors in both competitions and how her project has developed since last summer.
How does it feel to be a winner at Intel and at Google?
Being a winner at Google and Intel is very surreal. Having top scientists from both competitions recognize me as a top competitor provides me with an amazing affirmation about my STEM future. It is also exciting because both competitions assess different elements. Google was very driven by my research where as Intel STS focused on evaluating my potential as a scientist. At Google, I presented my research and was questioned on that. At Intel, questions could range from “Evolutionarily speaking, could dragons exist?” to more concrete questions about complex science concepts.
Did you change anything about your project since Google?
The Google Science Fair has given me a platform to share my research with the world. Since the Google Science Fair, I have been collecting more data to improve my breast cancer program. So far, the network has diagnosed 100% of the additional samples correctly, and the exciting part about collecting more samples is not only am I validating my research, but also through 7.6 million trials I’ve proven that these additional samples will also improve the accuracy of my program. Since my program learns based on experience, the more experience, the better. In addition, I have built a REST (representational state transfer) service so that an institute in Italy can test the program against 400 dubious samples. [REST is a program that facilitates transactions between Web servers].
In addition, I extended my research to diagnose MLL leukemia samples from genetic expression profiles. A hybrid neural network is 100% accurate and was also able to identify four genetic expressions that are particularly important in the decision. These regions may be places for drug companies to target since MLL is an aggressive form of leukemia with no good treatment option.
Because leukemia and breast cancer are so different (leukemia is a blood cancer with 12,582 genetic expressions as inputs where as breast cancer deals with masses and 9 cytological inputs), I am hopeful that my neural network technique will be applicable to all types of cancer diagnostics.
Any other updates on your project — have you gotten any doctors or hospitals to start using it, are you going to write up the research for publication?
I have started working with Lankenau Medical Center and an Institute in Italy (see above). I also aspire to get my research published.
Have your career goals changed since your GSF win and your internship?
This year has been amazing! It has confirmed my desire to pursue a computer science degree along with a biology degree, and I still want to become a pediatric oncologist. Most likely, I would like to earn an Md-PhD since I want to be able to contribute to a team that is working to find the cures to cancers. However, I would at some point love to work/intern for Google since I have found it to be an amazing company.
I think overall this year has shown me how many people care about me and my research, and how much support is out there if I just ask.