E-Health Tools For Melanoma Screening

Where We Are and How We Can Improve

Ashley Crew, MD
Assistant Professor of Dermatology
Keck School of Medicine of the
University of Southern California

Abhilashi Tyagi, BS
Keck School of Medicine of the
University of Southern California

Kimberly Miller, MPH
Department of Preventive Medicine
Keck School of Medicine of the
University of Southern California 

Myles Cockburn, PhD
Associate Professor of
Preventive Medicine and Dermatology
Keck School of Medicine of the
University of Southern California

Acknowledgments: This work was supported in part by the National Cancer Institute and the National Institute of Child Health and Human Development under grant RO1CA158407 to Dr. Cockburn.

According to the World Health Organization, “e-Health” encompasses three primary areas: (1) the delivery of health information by electronic means (2) the use of information technology (IT) and e-commerce to improve public health services; and (3) the use of e-commerce and e-business practices in health systems management.1 Whether or not the term “e-Health” is regularly employed in dermatology vocabulary, the different approaches to e-Health stand to affect the practice of dermatology significantly in the coming years. Its initiatives can be fundamental, such as the use of electronic medical records (EMR) in a growing number of practices. The implementation of EMR requires the assessment and incorporation of new behaviors into daily schedules.

But e-Health can also be introduced through subtler tools—such as the widely available applications designed to evaluate lesions for risk of melanoma and other skin cancers. Physicians have the option of simply observing the use and dissemination of these new apps, or more scrupulously assessing and critically evaluating their use.

The development of mobile apps for user-dependent secondary prevention of melanoma is exactly where, we would argue, dermatologists must take an active role in evaluation and assessment. In 2012, we published a catalog of available e-Health tools, highlighting what had to date undergone limited evaluation and assessment.2 Since publication of that review, the U.S. Food and Drug Administration (FDA) has recognized the functionality, rapid pace of innovation, and potential benefits of mobile apps, along with the risks they might pose to public health.3 Accordingly, in 2013, it issued a guidance on mobile medical applications (Mobile Medical Applications —Guidance for Industry and Food and Drug Administration Staff) announcing its intention to apply its regulatory authority to certain apps that could pose a risk to patient safety if not functioning as intended.

Here, we provide an update on mobile apps for secondary prevention (screening) of melanoma that are currently available through mobile platforms and the Internet.4 The applications discussed will include educational tools for the general population and those for physicians, algorithm analysis applications, “track-and-remind” applications, and “store-and-forward” applications. Examples of each will be included—how these tools work, what their benefits and limitations are, and how they could be used effectively by patients and physicians to improve melanoma screening. We review the most commonly used (or most widely available) of each set of tools, then summarize efforts made to date to evaluate their use in secondary prevention of melanoma and other skin cancers. Key areas for evaluation include sensitivity and specificity of the tool compared to a gold standard; repeatability or reproducibility; and whether or not the tool is likely to be widely translatable, based on the behavioral theory guiding its development. Finally, we summarize the general benefits and limitations of e-Health tools and each category of tool in Tables 1 and 2, respectively. Figure 1 summarizes the patterns of use among patients and physicians.

Table 1. Benefits and Limitations of e-Health Tools

Educational Applications for the General Public

Several applications are designed to teach proper skin self-exam techniques and instruct users in early recognition criteria such as the ABCDEs of melanoma. The delivery of information varies from application to application, but often involves a combination of expository text and diagrams, demonstrative videos, or interactive features such as quizzes. An example of this type of app is My MoleChecker, which has an embedded video teaching the ABCDEs of melanoma, along with a gallery of sample photographs of potentially dangerous moles.5 This app also includes a mole-tracking and reminder function.

The potential benefits of such educational apps are significant. For many individuals, the information in them may be more convenient and accessible than education provided by a busy clinician. The digital format allows incorporation of diagrams or demonstrations, which may not be readily available in a clinician’s office. These educational apps are also cost-effective compared to the education that a patient could receive in a clinician’s office. The cost of these apps for the health care industry is limited to the cost of production and distribution. For consumers, many of these apps are available as free downloads to mobile devices.

Educational Applications for Clinicians

The educational tools for physicians are similar to those targeted to the general public in that the goals are to increase the practice of proper skin cancer examination and improve the user’s ability to recognize potentially dangerous lesions.

The Integrated Skin Exam is a film available through the American Academy of Dermatology website that describes appropriate clinical technique and reviews the ABCDE criteria for melanoma.6 This film targets medical students, addressing their knowledge gaps in skin cancer examination, and has been used as an educational intervention in several medical schools.7 After watching the film, medical students improved their knowledge of melanoma, demonstrated increased confidence in skin cancer examination (66.93 percent vs 16.40 percent, P< .001), and had augmented intentions to practice it (99.05 vs 13.9 percent, P<.001).7

Dermoscopy Two-Step Algorithm is a mobile app targeted toward health care professionals, particularly those who may routinely use dermoscopy, such as a resident in a dermatology training program.8 The application describes various dermatoscopic patterns of benign nevi compared with melanoma-specific structures and provides sample photographs of these patterns. The app additionally features several cases that allow users to test their knowledge and answer quiz questions.

The value of these educational applications is that they both address knowledge of skin cancer examination as a practice gap in medical training,7 and provide easily accessible reminder references for clinicians who have already been trained in skin examination and the features of melanomas.

These physician apps have the same potential for misunderstanding and misapplication of information that exists in the apps for the general public. We recommend using them in the context of a training program, e.g., medical school or a residency training program, in which users have the opportunity to clarify ambiguities and address questions to experienced clinicians.

An additional limitation of physician apps that only describe the visual appearance of lesions is that they may underemphasize the importance of patient history and clinical context in guiding treatment decisions about whether to biopsy, reassure the patient, or closely monitor skin lesions.

Evaluation of Educational Tool Applications

There is relatively little oversight in the development and use of applications targeted towards patient populations. Therefore, the potential for misinformation or misapplication of information is high. Because of this, it is imperative that educational applications go through a rigorous process to assess whether the practices they advocate are in line with the standards of screening and general teaching of the American Academy of Dermatology (AAD) and the American Cancer Society (ACS). One option for oversight would be to create a committee through the AAD that would be charged with developing criteria to assess accuracy. Programs approved through this process would be awarded a certificate of compliance that would ensure patient and physician consumers of their safety and utility. It would also be ideal to incorporate use of these applications within the context of a physician-patient relationship, whereby the physician could recommend a specific app and then follow up with the patient later to review the information and answer any questions.

Although health care providers, given their prior training, are presumably a less vulnerable population than the lay public, it remains important for the physician educational apps to be assessed based on their ability to improve knowledge and change behavior. They must also be evaluated to ensure that their recommendations and teachings are in line with the recommendations and principles of the AAD and ACS.

Algorithm Analysis Applications

Algorithm analysis apps are designed to analyze user-uploaded photographs of skin lesions and provide risk profiles. Since the FDA issued its guidance on mobile medical applications in 2013, nearly two years ago, many of the algorithm analysis apps available then are no longer available for download from Apple and Android App stores.2 Skin of Mine and Doctor Mole9 are two apps we were able to find for download. Skin of Mine has a feature that allows users to take photographs of moles. The app can then perform an automated analysis of warning signs appearing in the photograph, including asymmetry, border scallop, border fade, and color irregularity.10 Although each algorithm varies, some commonly used ones are based on approximating dermoscopic clinical pattern analysis using the ABCD criteria,11 the 7-point checklist,12 or Menzies criteria.13 These applications depend on users identifying which lesions should be monitored.

In the right context, the potential utility of such applications is significant. A clinical device known as MelaFind, designed to aid dermatologists in deciding whether or not to biopsy, was shown in a recent study to have an estimated biopsy sensitivity of 96 percent with a 95 percent lower confidence bound at 83 percent.14

Human Error

However, applications targeted directly to patients are concerning because they lack sensitivity and depend on an untrained individual to identify suspicious lesions. A study that assessed the diagnostic accuracy of three mobile applications reported sensitivities ranging from 6.8 to 70 percent and specificities ranging from 37 to 93.7 percent.15 These applications have been controversial given that overconfidence in inaccurate risk assessments may lead to delays in seeking physician consultations when melanomas are still at a treatable stage. This is of particular concern for those with limited access to medical care who may use an application in lieu of physician consultations.

Even with very high diagnostic accuracy, these apps should not completely replace clinical skin exams; melanomas on body sites such as the scalp that are difficult to see might go unnoticed, and users might focus only on lesions that are concerning to them, overlooking clinically significant lesions. To our knowledge, no algorithm applications are able to assess certain malignant lesions such as atypical variants of melanoma (e.g., amelanotic melanoma) or other types of skin cancers such as squamous cell carcinoma or basal cell carcinoma, which would be detected in a clinical skin exam.

Evaluation of Algorithm Analysis Applications

In contrast to store-and-forward programs, algorithm analysis apps can be used to evaluate and prognosticate without the third party assistance of a dermatologist. But this lack of a specialist’s direct oversight increases concern about both false reassurance and unnecessary biopsy. It is thus imperative that the applications go through rigorous pre- and post-market testing to assess sensitivity and specificity.

Ideally, applications should be compared to the “gold standard”—a dermatologist’s exam followed by histologic evaluation of lesions suspected to be malignant. It is also imperative to evaluate the ability of the apps to assess quality of images and request improved images when necessary. Image assessment should be tested on varied skin types as well as various types of lesions, since it cannot be guaranteed that only melanocytic lesions would be submitted for assessment. The applications themselves should also be piloted on individuals with varied levels of proficiency with technology.

Algorithm analysis apps used by dermatologists have the obvious additional safety net of a dermatologist’s direct assessment. But because dermatologists presumably use the applications for assistance in diagnosis, it seems reasonable to assume that their diagnostic and therapeutic choices will be affected by these applications. If this is true, it is again crucial for the applications to be tested rigorously for sensitivity and specificity before being incorporated into a dermatologist’s daily practice. Although diagnostic accuracy may improve in the future with better camera quality and improved algorithms, the lack of extensive study of these applications and their significant potential for harm lead us to recommend very cautious use at this time.

Track-and-Remind Applications

Apps in this category employ mole-mapping devices that utilize the camera feature in smartphones and tablets to take photographs of skin lesions and track them over time. These applications also generally include a record of the date and time when the photograph was taken and store the lesion’s body location manually entered either in text or on a three-dimensional model of a body. Many of these apps also come with an option to set a periodic alert reminding the user to perform a subsequent skin exam.

One app, Mole Monitor,16 allows users to view side-by-side and overlay images to assess for lesion changes or evolution. It can also produce a mole record (a document including medical history questions related to skin cancer risk assessment and the photographs of a particular lesion over time), which can be presented to a clinician. The developers of this application optionally suggest the use of an attachable macro lens to improve the quality of close-up photographs.16 Unlike attachable mobile dermatoscopes, intended for use by clinicians,2 macro lenses are targeted to consumers for general photography purposes. They can also improve image quality for use with algorithm apps or teledermatology apps.

If used appropriately, these apps and their summary record may increase the accuracy and efficiency of a physician’s health care assessments. Additionally, a record of high-quality images of skin lesions with the use of a macro lens may be particularly useful for intermediate-risk lesions, for which close monitoring may be favored to reduce the number of preemptive biopsies.

Evaluation of Track-and-Remind Applications

Track-and-remind apps are less concerning than others from a public health standpoint, in that they primarily serve to remind patients to perform skin exams and help them track clinical images; they do not assess whether or not lesions are worrisome. They are therefore less likely to provide false reassurance to patients and prevent them from seeking medical care for a potentially dangerous lesion.

It is important for the behaviors advocated in these applications to be in line with current practices, and for the program’s limitations (e.g., inability to assess or recommend intervention) to be carefully spelled out to protect consumers. A committee similar to the one proposed to oversee educational applications could be used to assess track-and-remind apps.

Similar to the limitation of algorithm analysis apps, mole-tracking apps depend on users’ ability to identify clinically significant lesions. Moles that are difficult to photograph or recognize may be missed.

Because they allow patients to “track” concerning lesions, delaying assessment by a medical professional, we recommend that mole-tracking and reminder applications be used under the guidance of a supervising physician.

Teledermatology Store-and-Forward Applications

Teledermatology applications exist as a platform for users to communicate with a physician. They generally function on a pay-per-use schedule and allow users to send photographs of particular skin lesions for assessment by a clinician, often within a specified time frame such as 24 hours. Many of these apps also have users submit answers to personal history questions and allow them to ask specific questions about a lesion.

Although diagnostic accuracy may depend on the clinician who is reviewing images, a review of existing teledermatology literature found the concordance of store-and-forward teledermatology to be moderate to excellent. When management options were specifically listed (i.e., to biopsy or not to biopsy), concordance ranged from 72 to 96 percent.17

Teledermatology apps can increase access to care, particularly for individuals in medically or geographically underserved groups. These applications may also yield more efficient use of health care by triaging easily evaluable skin lesions.

Evaluation of Store-and-Forward Applications

Assuming that the physician on the receiving end of these applications is a board-certified dermatologist, much of the work to validate the use of these programs has already been performed. While it is reassuring that a clinician participates directly in the management of these cases, the concern exists that these apps would be viewed as an adequate replacement for total-body skin exam.

As with other applications that depend on user-uploaded photographs, patients using this application may ignore clinically significant lesions because they are difficult to spot or because the user is not familiar with the appearance of certain lesions. For this reason, store-and-forward applications are currently appropriate for consultations on specific lesions only. Ideally, this technology would be employed in a setting where patients have a pre-established pathway to access a clinician for lesions identified as potentially dangerous, as well as for routine total-body skin exams.

Because this application utilizes transmission of information to another party, there are also breach-of-privacy concerns.


The potential benefit of e-Health tools in melanoma detection derives largely from their ability to reach a wide-ranging population: widespread access to the Internet and smartphones allows for easy distribution of useful tools to a large population. Certain applications may also increase access to medical care, particularly for users in remote locations, in a way that is also cost-effective. While the potential positive impact is large, it is imperative to evaluate the accuracy of these applications clinically, using the same level of scientific rigor we would apply to any form of screening test. The e-Health tools highlighted here have some of the same possible drawbacks of any screening mechanism, and we should systematically evaluate the sensitivity and specificity of each tool against known gold standard techniques. While the widespread availability of the tools is a potential advantage in reaching at-risk populations, it also makes it difficult to track users to determine whether the benefits outweigh the potential harm. Perhaps the greatest risk is being given a false sense of security if the tool fails to identify a dangerous lesion. Furthermore, if there is no available follow-up for a positive finding, the tool will not be successful in secondary prevention.

Most screening-related e-Health tools rely on self-motivated individuals to complete their own skin examinations, yet there has been little research conducted on whether or not the tools successfully integrate what is known about motivations for healthy behaviors. To ensure that they align with patient needs and abilities, assessments should focus on patient motivation and barriers to use. An informed understanding of the theoretical and behavioral basis of dermatological e-Health tools will help ensure their adoption as useful, patient-centered applications that are likely to be successful in secondary prevention.

The availability of e-Health tools rapidly changes, reflecting improvements in technology and new innovations. However, this rapid change also highlights the need for frequent expert reviews and updated catalogs of these tools not only to identify potentially harmful apps but also to recommend applications that will be beneficial for patients and health care providers. Further studies on e-Health tools should describe implementation, effectiveness, distribution, and cost-analysis.


  1. World Health Organization Staff. E-Health. World Health Organization. http://www.who.int/trade/glossary/story021/en/. Accessed October 1, 2014.
  2. Tyagi A, Miller K, Cockburn M. e-Health tools for targeting and improving melanoma screening: a review. J Skin Cancer 2012; doi:10.1155/2012/437502. Epub 2012 Dec 13.
  3. Food and Drug Administration Staff. Mobile Medical Applications Guidance for Industry and Food and Drug Administration Staff. http://www.fda.gov/downloads/MedicalDevices/.../UCM263366.pdf. Issued September 25, 2013. Accessed September 24, 2014.
  4. Alpay L, van der Boog P, Dumaij A. An empowerment based approach to developing innovative e-health tools for self-management. Health Informatics Journal 2011; 17(4):247–255. Accessed November 10, 2014.
  5. Embarrassing Bodies My MoleChecker. 2012. Available at: http://itunes.apple.com/us/app/embarrassing-bodies-my-molechecker/id503167508?mt=8. Accessed August 15, 2014.
  6. Basic Dermatology Curriculum. Available at: http://www.aad.org/education/basic-dermatology-curriculum. Accessed August 15, 2014.
  7. Garg A, Wang J, Reddy SB, et al. The Integrated Skin Exam film: An educational intervention to promote early detection of melanoma by medical students. Journ Amer Acad Dermatol 2014; 70(1):115-119. Accessed November 10, 2014.
  8. Dermoscopy Two Step Algorithm. 2014. Available at https://itunes.apple.com/us/app/dermoscopy-two-step-algorithm/id731753300?mt=8. Accessed August 15, 2014.
  9. Doctor Mole—Skin Cancer App. 2014. Available at: https://itunes.apple.com/us/app/doctor-mole-skin-cancer-app/id562711395?mt=8. Accessed August 15, 2014.
  10. Skin of Mine. 2014. Available at: https://itunes.apple.com/us/app/skin-of-mine/id432628083?mt=8. Accessed August 15, 2014.
  11. Stolz W, Riemann A, Cognetta AB, Pillet L. Abcd rule of dermoscopy: a new practical method for early recognition of malignant melanoma. Eur J Dermatol 1994; 4:521-527.
  12. Argenziano G, Fabbrocini G, Carli P, et al. Epiluminescence microscopy for the diagnosis of doubtful melanocytic skin lesions. Arch Dermatol 1998; 134:1563-70. Accessed November 10, 2014. 
  13. Menzies SW, Ingvar C, Crotty KA, McCarthy WH. Frequency and morphologic characteristics of invasive melanomas lacking specific surface microscopic features. Arch Dermatol 1996; 132(10):1178–1182. Accessed November 10, 2014.
  14. Wells R, Gutkowicz-Krusin D, Veledar E, et al. Comparison of diagnostic and management sensitivity to melanoma between dermatologists and MelaFind: a pilot study. Arch Dermatol 2012; 148(9):1083-1084. doi:10.1001/archdermatol.2012.946. Accessed November 10, 2014.
  15. Wolf JA, Moreau JF, Akilov O, et al. Diagnostic inaccuracy of smartphone applications for melanoma detection. JAMA Dermatology 2013; 149(4):422-426. Accessed November 10, 2014.
  16. Mole Monitor. 2014. Available at: http://www.mole-monitor.com/. Accessed August 15, 2014.
  17. Warshaw E, Greer N, Hillman Y, et al. Teledermatology for diagnosis and management of skin conditions: a systematic review of the evidence. VA-ESP Project #09-009; 2009. http://www.ncbi.nlm.nih.gov/books/NBK49157/. Accessed November 10, 2014.