Pathologists take heed! Teenagers are taking off-the-self technology and creating useful new clinical laboratory tests for cancer and other diseases
For the second time in recent weeks, a teenager has made national news for developing a medical laboratory test that can more accurately diagnose disease when compared to methodologies currently used by clinical laboratories and pathology groups. Pathologists and clinical chemists have good reason to ask what is different about the science taught in today’s high school compared to recent years.
The subject of our coverage today is a 17-year-old girl from Sarasota, Florida. She developed a computer application that detects breast cancer with 99% accuracy! It was on July 6 when Dark Daily e-briefing, told you about the 15-year old from Maryland who developed a diagnostic test to detect early-stage pancreatic cancer. (See “High School Student Develops Diagnostic Test to Detect Early-Stage Pancreatic Cancer”.)
A story published at Foxnews.com tells how Brittany Wenger won first-place in the second annual Google Science Fair for her elegant computer application that increases to 99% the accuracy of fine-needle aspiration (FNA) diagnostic testing for breast cancer.
Currently, FNA is the least invasive diagnostic test for the disease. It is also often the least conclusive. In these circumstances, patients sometimes have to undergo a second FNA with a larger needle, or even, in some cases, have a surgical biopsy, FoxNews reported.
Teen-ager Brittany Wenger of Sarasota, Florida, developed a software program that was able to more accurately diagnose breast cancer from specimens collected by fine needle aspiration (FNA). Her effort won second place at the Google Science Fair, along with national news coverage of her accomplishment. (Photo by Andrew Federman of Google.)
“This program aids doctors in the detection process to minimize inconclusive and misdiagnosed samples,” Wenger wrote in the slides she submitted to the competition.
Wenger’s software application is based on an artificial neural network (ANN). ANNs are information processing systems modeled on the human brain. These networks are “trained” to detect complex relationships and patterns.
In Wenger’s application, the networks are shown examples of “good” and “bad” patterns, explained an article titled “Neural Networks & Genetic Algorithm” and published on eHow.com. The internal statistical weights of the network are adjusted slightly every time the program identifies a pattern incorrectly. As more samples are input, the network learns to identify all patterns correctly.
Teen’s Breast Cancer Test Could Reduce More Invasive Testing
As anatomic pathologists are aware, most women who traditionally are referred for a biopsy do not have cancer. According to one study published in the Journal of the American College of Surgeons, of the cases reviewed, only 23% of the women biopsied for breast cancer were subsequently diagnosed with the disease.
“Exposing large numbers of women who do not have cancer to invasive procedures may be considered an undesirable medical practice,” the report authors wrote. “In conclusion, many women could benefit from highly accurate noninvasive tests.”
“I have had a lot of family members and a lot of family friends who have had breast cancer,” Wenger stated during an abc Action News interview. According to the American Cancer Society website, breast cancer is the second leading cause of cancer death in women. This motivated the teen to develop an ANN to boost the success rates of the less invasive FNA diagnostic method.
Teen’s Neural Network Outperformed Commercially Available Networks
“Neural networks can be applied to medical diagnostics and outliers will be identified by modern technology,” Wenger hypothesized in her prize-winning project. She tested three commercially available ANNs. Also, she tested an ANN that she programmed herself, FoxNews reported.
Wenger discovered that the network that she had programmed was the most reliable. She correctly identified cancerous samples with 99% accuracy. Her program yielded “inconclusive” results about 4% of the time. It resulted in false negatives less than 1% of the time, compared to a 5% false negative rate with commercial neural networks, the FoxNews story stated.
Wenger fed publicly-available data from FNAs of breast cancer patients into several different ANNs. Each of the neural networks “learned” how to diagnose breast cancer by analyzing the data and recognizing various characteristics of the different samples.
“I think [the application] might be hospital ready,” Wenger stated in the ABC interview.
Wenger beat out thousands of entries from 100 countries to win the grand prize: an internship at one of the institutions hosting the science fair, a trip to the Galapagos Islands, a trophy made of white Legos, and a $50,000 college scholarship.
Wenger plans to major in computer science in college and work as a pediatric oncologist, she told ABC. “With more samples, the network may achieve perfection,” the teen stated in her project summary. “I’d love to get different data from doctors,” she added in a Huffington Post story.
The ability of this young scientist to develop a breakthrough diagnostic application for breast cancer diagnosis demonstrates the wide-scale accessibility of sophisticated and easy-to-use information technology. As diagnostic technologies continue to evolve, pathologists and clinical laboratory managers can expect to see ever more examples of similar diagnostic discoveries developed by unexpected sources.
—Pamela Scherer McLeod
Related Information:
High School Student Develops Diagnostic Test to Detect Early-Stage Pancreatic Cancer
17-year-old programs artificial ‘brain’ to diagnose breast cancer
Brittany Wenger, 17, Wins Google Science Fair Grand Prize For Breast Cancer Diagnosis App
