Brittany Wenger Wins Google Science Fair for Programming
Seventeen-year-old Brittany Wenger has won the 2nd Annual Google Science Fair for her work in developing an application to help doctors diagnose breast cancer!
The Google Science Fair is a science talent competition for kids ages 13 to 18, held this month in Palo Alto, California. Brittany Wenger developed a cloud-based computer program that makes breast cancer detection less invasive “ the œGlobal Neural Network Cloud Service for Breast Cancer. She coded her applications to think like the human brain and then used them to locate mass malignancy in breast tissue samples.
This method is a great deal less taxing than theminimally invasive, but painful, biopsy called a fine needle aspirate (FNA). Wegner tested her method with 7.6 million trials to see how accurately it would detect cancerous tumors... to a astonishingly97.4 percent success rate in prediction and 99.1 percent sensitivity to malignancy.
What does this mean? It means that the genius of this young programmer gives us a tool for potentially empowering doctors and patients through a less invasive (but highly effective) method of assessing tumors.
Previously, non-invasive breast cancer diagnosis had not been nearly as successful as invasive procedures. However, Brittany's idea could help move the field forward. Her computer program, called a ""neural network,"" mimics the human brain in that it reads massive amounts of information and detects complex patterns, and then can ""learn"" to make diagnostic calls on breast cancer.
For winning the competition, Wegner received $50,000, a trip to the Galapagos Islands and one year of mentoring and internship opportunities. As for her future, Wegner said in a recent interview that she plans to major in computer science in college and attend medical school.
What first inspired her to look into neural networks was one of her courses, an elective centered on futuristic thinking. ""It was really by accident that I discovered this amazing technology and became so enthralled,"" Wenger says. ""When I came across it, I went out and bought a programming book, and with no experience decided that that was what I was going to do.""
Wenger noticed that the results from one of the least invasive biopsies, the fine-needle aspiration, were the most inconclusive. She worked to design a system that would analyze the samples more intelligently. ""I created a tool for doctors to use that could detect patterns in these fine-needle aspirates that are too complex for humans to detect so that it could provide doctors with this way to determine whether masses are malignant or benign without all the invasion,"" Wenger says.
The network takes into account qualitative data that help identify whether or not the masses are malignant or benign, then makes connections between the different inputs to produce an output. ""In training mode, is going to know the answer it should have gotten, and it's going to try to adjust itself so that it can get that answer next time,"" Wenger says. ""In testing mode, it applies what it's learned.""
The program has a success rate of 99.11 percent of correctly identifying malignant and benign tumors, and the high school senior points out that that figure will only improve. The program will increase the effectiveness of the fine-needle aspiration test, which is good news for people being diagnosed.
Audio excerpt (from The Takeaway linked above)
Brittany Wenger Wins Google Science Fair for Programming
Seventeen-year-old Brittany Wenger has won the 2nd Annual Google Science Fair for her work in developing an application to help doctors diagnose breast cancer!
The Google Science Fair is a science talent competition for kids ages 13 to 18, held this month in Palo Alto, California. Brittany Wenger developed a cloud-based computer program that makes breast cancer detection less invasive “ the œGlobal Neural Network Cloud Service for Breast Cancer. She coded her applications to think like the human brain and then used them to locate mass malignancy in breast tissue samples.
This method is a great deal less taxing than theminimally invasive, but painful, biopsy called a fine needle aspirate (FNA). Wegner tested her method with 7.6 million trials to see how accurately it would detect cancerous tumors... to a astonishingly97.4 percent success rate in prediction and 99.1 percent sensitivity to malignancy.
What does this mean? It means that the genius of this young programmer gives us a tool for potentially empowering doctors and patients through a less invasive (but highly effective) method of assessing tumors.
Previously, non-invasive breast cancer diagnosis had not been nearly as successful as invasive procedures. However, Brittany's idea could help move the field forward. Her computer program, called a ""neural network,"" mimics the human brain in that it reads massive amounts of information and detects complex patterns, and then can ""learn"" to make diagnostic calls on breast cancer.
For winning the competition, Wegner received $50,000, a trip to the Galapagos Islands and one year of mentoring and internship opportunities. As for her future, Wegner said in a recent interview that she plans to major in computer science in college and attend medical school.
What first inspired her to look into neural networks was one of her courses, an elective centered on futuristic thinking. ""It was really by accident that I discovered this amazing technology and became so enthralled,"" Wenger says. ""When I came across it, I went out and bought a programming book, and with no experience decided that that was what I was going to do.""
Wenger noticed that the results from one of the least invasive biopsies, the fine-needle aspiration, were the most inconclusive. She worked to design a system that would analyze the samples more intelligently. ""I created a tool for doctors to use that could detect patterns in these fine-needle aspirates that are too complex for humans to detect so that it could provide doctors with this way to determine whether masses are malignant or benign without all the invasion,"" Wenger says.
The network takes into account qualitative data that help identify whether or not the masses are malignant or benign, then makes connections between the different inputs to produce an output. ""In training mode, is going to know the answer it should have gotten, and it's going to try to adjust itself so that it can get that answer next time,"" Wenger says. ""In testing mode, it applies what it's learned.""
The program has a success rate of 99.11 percent of correctly identifying malignant and benign tumors, and the high school senior points out that that figure will only improve. The program will increase the effectiveness of the fine-needle aspiration test, which is good news for people being diagnosed.
Audio excerpt (from The Takeaway linked above)