If you’re worried that a questionable mole might be skin cancer, there’s an app for that — thanks to Emma Nordstrom, a St. Charles volunteer and graduate of Bend’s Trinity Lutheran School.
Earlier this year, Nordstrom, 18, was one of 200 high school students in the country to be recognized as a National STEM (Science, Technology, Engineering and Math) Champion for her invention: A program that allows a person to scan a concerning lesion and will flag it if it shares characteristics with different types of skin cancer.
Nordstrom — who spends Monday afternoons visiting and brightening the days of St. Charles patients — initially created the program as a project for the International Science and Engineering Fair.
“I decided to go down the path of skin cancer, particularly because underrepresented communities are often more affected by skin cancer,” she said.
As a sophomore, Nordstrom developed the program to run on a computer using a type of artificial intelligence called machine learning, specifically a convolutional neural network. That means the AI is trained in a way that mimics the human brain. Just as children learn over time that a square shape with legs is a table, the neural network learned from thousands of images to decipher what is and what isn’t skin cancer.
Nordstrom used around 10,000 images of skin cancer to train the program. Those images covered three forms of the disease: Basal cell carcinoma, squamous cell carcinoma and melanoma. (A fourth set of images, benign lesions, was used as a control.)
The program will flag a lesion as suspicious if it matches skin cancer images used in training. It’s not meant to be a diagnostic tool, but rather a screening tool to give people guidance as to whether they should see a dermatologist. Primarily, Nordstrom envisions physicians using the program to help them decide whether to refer patients to a specialist. While wait times to see a dermatologist can stretch beyond six months, a physician who suspects a lesion might be skin cancer can fast-track the referral.
Through the development process, Nordstrom decided to move the program from a computer to a smartphone.
“Not everyone has access to a computer,” she said. “We might be able to use this here in Bend because we are a higher income area, but I want to make sure it’s accessible in lower income areas, too.”
As a junior, Nordstrom competed at ISEF with the app version of the program, called DermaSkan. Then, as a senior, she submitted it to the National STEM Challenge, a nationwide competition in which students submit a science-related project that attempts to overcome a real-world problem.
Roughly 4,000 people submitted projects to the STEM challenge. Nordstrom was one of fewer than 200 National STEM Champions.
“I was shocked,” she said.
Kara Magee was not shocked. She’s the coordinator of volunteer services for St. Charles, and a huge fan of Nordstrom, whom she met just over a year ago.
“When she first arrived here, she was telling me about this app,” Magee said. “I was like, ‘You are going to save people’s lives!’ We’re all one of a kind, but she is just an exceptional human being.”
Although Nordstrom hopes to obtain a patent for her app in the future, she isn’t planning to pursue medical AI as a career, she said. Instead, she will attend Oregon State University next fall, where she’ll start down the path of becoming a pediatric nurse.
That’s no surprise to Magee.
“She goes in and chats with patients every week, which is one of the most important jobs we have here because some of them have no one to talk to,” she said. “She’s already making a difference in their lives, and she’s going to make a huge difference in our world.”
