CANCER CALCULATIONS US teen’s computer algorithms improve cancer diagnosis accuracy
While some teenage girls spend their days obsessing over the latest movie release or boy band, Brittany Wenger pursues a very different passion – diagnosing, and perhaps curing, cancer with complex computer programs she created after teaching herself to write the code.
Using artificial neural networks – computer programs coded to “think” and “learn” like the human brain – 18-year-old Sarasota, Florida (USA) student and self-taught computer coder Brittany Wenger has written a program that diagnoses breast cancer and an aggressive form of childhood cancer called mixed-lineage leukemia (MLL) more accurately than ever before possible (see related article on Page 38).
Wenger’s research, which earned her the prestigious 2012 Google Science Fair Grand Prize, the Intel ISEF Grand Award and a trip to the White House to present her work to US President Barack Obama, has gone viral, prompting oncologists worldwide to contribute their samples to her database, which has raised its predictive accuracy to 99.11% and climbing.
“The whole judging panel came away with a big ‘wow,’” said Vint Cerf, widely acknowledged as one of “the fathers of the Internet” and the vice president and chief Internet evangelist at Google.
Wenger’s work also has identified small subsets of proteins (20 proteins in the case of breast cancer and just four in MLL) from the tens of thousands involved in each disease that are the best predictors of which cells will prove cancerous, information that should help pharmaceutical companies hone in on a short list of potential new drug targets.
“Because artificial neural networks aren’t told exactly how to do a specific task, they learn based on experiences and mistakes,” Wenger explained. “And since they have this learning quality, they can be ‘trained’ to recognize patterns that are far too complex for humans to detect. The implications of this pattern-recognition capacity are very exciting in every field of science.”
The program becomes increasingly accurate as it “learns” from more and more data points, so Wenger was adamant about putting her network on the Internet connected cloud. “The cloud is able to connect the world and researchers like never before,” Wenger said. “I’m able to work with a hospital in Italy, for instance, through a web service. They’re able to provide input into my program with different samples, conveniently, remotely and inexpensively. Having that power is huge.”
Wenger’s programs are being beta-tested in hospitals in the US and Italy, and she is further customizing the network for use with ovarian and lung cancer.
Perhaps the only thing more impressive than Wenger’s extensive list of secondary school accomplishments may be the promise of her future. In August 2013, she entered Duke University in Durham, North Carolina (USA), as one of eight Angier B. Duke Scholars in the class of 2017, with plans to pursue a career in pediatric oncology. The Duke financial awards, which pay the full cost of four years at the top-tier university, are among the world’s most competitive scholarships.
Although always an inquisitive child with an interest in science, Wenger’s passion for the crossroads between computing and medical research was inspired by a cousin’s battle with breast cancer and fueled by a collection of dedicated teachers and the unwavering support of her parents. “I was the kid who never outgrew the ’why?’ phase,” she said. “A lot of parents probably would have put their foot down when I was setting my alarm clock every three to four hours to test some of the programs I’ve created, but my parents have always understood how important my research is to me.”
And now, thanks to that support, Wenger’s research is offering new hope to cancer patients worldwide and new support for researchers determined to defeat the disease. ◆