Tinder-style app helps medical students identify melanoma
A smartphone app may help medical students more accurately differentiate melanomas from benign skin lesions, according to study results published in JAMA Dermatology.
The traditional method of skin cancer detection is based on ABCDE criteria. The checklist accounts for asymmetry, border irregularity, color, diameter and evolution of a lesion.
Relying solely on these criteria and textbook examples does not provide students with the necessary tools to develop an intuitive eye for malignant lesion detection, according to Michael Kolodney, MD, PhD, chair of West Virginia University’s dermatology department.
Kolodney developed an app called Skinder to help medical students become better at identifying malignant lesions. Skinder works similarly to Tinder, a dating app on which users “swipe right” if they want to make a match and “swipe left” if they don’t.
With Skinder, users view a series of high-resolution images of skin lesions. Users swipe right if they think lesions are benign and swipe left if they think lesions are malignant.
Kolodney and colleagues assessed the app in a randomized diagnostic study that included 36 medical students at West Virginia University who had no formal clinical dermatology rotation. Researchers randomly assigned participants to use Skinder or follow a traditional diagnostic approach based on the publicly available Internet Curriculum for Melanoma Early Detection (INFORMED) Skin Education Series.
Each participant completed a 32-image pretest in which they were asked whether lesions were malignant or benign. Participants then had 1 hour of observed training time dedicated to their assigned learning modality.
After training, participants completed a posttest that included the same 32 images in a different order.
On the pretest, the mean percentage of correctly identified lesions was 74.7% in the INFORMED group and 75% in the Skinder group. On the posttest, the mean percentage of correctly identified lesions was 77.5% in the INFORMED group and 86.3% in the Skinder group (absolute difference, 9%; P = .003).
HemOnc Today spoke with Kolodney about the app, the efficacy it has demonstrated so far, and the potential benefits it could offer for melanoma detection.
Question: Why did you decide to develop this app?
Answer: Medical students get very little training in dermatology because there is a great competition for their time among all specialties. Dermatology often gets the short end of that. When it comes to recognizing things visually, people tend to learn better by experience than by memorizing a set of rules. We wanted to help students develop an intuitive sense of what looks malignant and what looks benign to help them better differentiate melanomas from other pigmented skin lesions. Medical students may see a handful of lesions in a textbook or during a lecture but, to get an intuitive sense of what is malignant or benign, you need to see thousands of them. We tried to develop an app that would expose them to a large number of lesions in a short time, while using a format that students would be comfortable with and would command their attention more than a lecture or reviewing an atlas of skin lesions.
Q: How do you hope Skinder will be used?
A : We hope to distribute the app free to whoever wants it. Our primary target is medical students, but it doesn’t take a background in medicine or science to use this app. It’s all just visual recognition. Users could include physicians, nurses, other medical providers or even patients.
Q: What benefits could the app offer to patients?
A: Oftentimes a patient’s spouse will spot the lesion that prompts someone to go to their doctor. People who have had one melanoma are very likely to get another in their lifetime. Many of those individuals are interested in surveillance and want to identify subsequent melanomas at the earliest possible stage. The app could have use for these individuals and their family members if they are motivated to learn it.
Q: What is the next step in testing?
A: The study we published was based on how participants improved immediately following a single exposure to the app. Now we’re interested in more of a real-world scenario in which the user would use the app periodically. You don’t learn that much from a 1-hour exposure to anything, so we want to determine whether users retain this knowledge a month later. If they don’t, it’s not terribly useful.
Q: Is anyone using Skinder now?
A: Access is limited right now. In some cases, we have permission to use the images to educate medical students but not to freely distribute them. Also, because we are doing more studies on the app, having more students using it could corrupt those studies. We are working on getting a more professionally created app. We also are asking physicians to donate their personal collections of images in hopes of getting access to thousands of additional images to distribute.
Q: How will you promote the app upon its release?
A: We’ll probably rely on word of mouth. Medical students use social media and talk to each other a lot so, if students derive benefit from it, word would spread pretty quickly among that group.
Q: How useful do you think Skinder can be for detection of cancerous skin lesions?
A: I think it has use in any situation when someone has to diagnose something visually, especially when they are making binary decisions. It could be used in a variety of other situations, such as radiology or ophthalmology. People learn best in different ways. I learn best from experience. Rather than staring at one image for 15 minutes, I’d prefer to look at a few hundred images in that same time and then repeat the process. When I talk to dermatologists, I ask them if they look at a lesion and think about the established rules. Almost everyone has said no. They tell me they look at it and know what it is based on their experience. I’m trying to simulate that experience in a short period of time for people who do not have access to a lot of patients.
Q: Is there anything else you would like to mention?
A: If physicians have collections of pigmented images and would like to share them for the app, we would be more than happy to work with them. The more images we have, the more powerful the tool will be. – by Scott Buzby
For more information:
Michael S. Kolodney, MD, PhD, can be reached at Section of Dermatology, Department of Medicine, West Virginia University, Fourth Floor HSCN, Room 4075-A, P.O. Box 9158, Morgantown, WV 26506; email: email@example.com.
Kolodney reports a grant from American Academy of Dermatology to fund the development of Skinder.