Artificial Intelligence may be making its way to your doctor’s office.
Commenting on recent research findings from a team at Stanford University, Sancy Leachman, M.D., Ph.D., director of the OHSU Knight Cancer Institute’s Melanoma Research Program and chair of the Department of Dermatology in the OHSU School of Medicine, co-authored a Nature News and Views perspective that heralded artificial intelligence, or AI, an important step in the “final frontier” in cancer diagnosis.
“They are close to having a computer with artificial intelligence that performs as well as board-certified dermatologists in its ability to discriminate melanomas from moles using digital images,” said Leachman. “If the technology lives up to its potential, users that need to be seen can get into their physician more quickly and those who have nothing of concern can avoid making an unnecessary trip to the doctor.”
Leachman is passionate about putting this type of technology in consumers’ hands so they can be proactive about their skin health. But artificial intelligence must be trained on large sets of relevant data, and scientists have not had access to patient-generated skin images and data. That’s why Leachman joined the design team for the mobile phone app Mole Mapper, which debuted to the public on the iTunes store in October 2015.
Mole Mapper was designed to allow people the ability to accurately measure and monitor their moles using their smartphone cameras. The app’s optional data collection function utilizes Apple’s ResearchKit platform and gives users the option to contribute to melanoma research by sharing anonymous photos of how their potential trouble spots evolve over time.
Today OHSU and Sage Bionetworks published data in Nature Scientific Data collected during the Mole Mapper app’s first seven months. By sharing this data, research teams with Institutional Review Board, or IRB, oversight can access this user-generated data.
What did researchers learn from the initial Mole Mapper data set?
Widespread participation is key — The study cohort includes 2,798 U.S. participants who contributed 1,920 demographic surveys, 3,274 mole measurements and 2,422 curated mole images.
Redheads are at increased risk for melanoma — “One of our most noteworthy findings confirmed that Mole Mapper participants with red hair were significantly more likely to be diagnosed with melanoma,” said Dan Webster, Ph.D., research fellow at the National Cancer Institute. “This is important because it confirms previously published data showing that individuals with red hair due to mutations in the MC1R gene have a higher risk for melanoma.”
User-generated data play an important role — “Artificial Intelligence, such as the kind used in the recent Stanford study, has an insatiable appetite for data to improve its results,” said Mole Mapper product manager and paper author Tracy Petrie, Ph.D. “Data is historically expensive to acquire, but the Mole Mapper project demonstrates how pairing mobile phone technologies with individual participation can help with that. It's really very exciting.”
It’s crucial to move away from data silos — Data collected from traditional research studies tend to be controlled by individual study researchers, creating disconnected data “silos.” The Mole Mapper study contributes to an effort by Sage Bionetworks to cultivate a more open mobile health ecosystem where data are shared broadly with qualified researchers worldwide. “It is truly a credit to the study participants who have agreed to broadly share these data with the research community,” says study senior author Brian M. Bot. “We as researchers owe it to the participants in these studies to reimagine how data are collected and shared in an effort to accelerate the research process.”