Artificial Intelligence in Dermatology Apps: A Comprehensive Review

Author: Dr. Alexander Börve

Date: 10/04/2024

How is AI Used in Dermatology Apps?

In simple terms, AI is a branch of computer science that aims to mimic human thinking and the ways we analyze complex data.1 Machine learning is a component of AI that allows computer programs to learn from experience without explicit instructions.2 In dermatology apps this feature is utilized extensively. Most AI algorithms are trained by using a vast collection of data; pictures and details of different skin conditions.

For years, researchers have been investigating how AI can enhance screening methods for skin cancer. Nasr-Esfahani et al. were the first to develop a neural network trained to detect melanoma, achieving sensitivity and specificity rates of 0.81 and 0.80, respectively.3 Since then, AI has been applied in various aspects of dermatology, including detection of other skin cancers, eczema, and psoriasis.4

However, despite the claims of all the potential benefits, lack of proper evidence to back them up makes us skeptical about using these apps. A recent study conducted by S. Wongvibulsin et al. comprehensively evaluates publicly available dermatology apps to present a scoping review of all the pros and cons they bring to the table.5

Review Method and Key Findings

The research methodology involved searching Apple and Android app stores from November to December 2023 using specific terms related to dermatology. Two investigators conducted the search, with discrepancies resolved by a third. Initially, 909 apps were identified, reduced to 41 after removing duplicates and non-medical or non-AI apps. This remainder was evaluated for each app's purpose, audience, evidence, algorithm details, data availability, clinician input, and privacy policies. Findings revealed concerns including lack of food and drug authority (FDA) approval, limited/absent disclaimers about the lack of approval, inadequate evidence, clinician input, and transparency regarding algorithm development and user privacy.

Understanding the Conclusions of the Study

Benefits revealed during the study include the ability to identify a range of skin conditions including skin cancer, diagnose skin conditions like acne, tracking the changes of moles, visual search options, and management of some conditions like acne, atopic dermatitis and eczema.5

The review concluded that while AI dermatology apps offer promising potential for improving access to care and patient outcomes, they also present significant risks. These risks primarily stem from the lack of consistent validation processes. Many of these apps lack robust validation procedures, which raises concerns about the accuracy and reliability of their diagnoses and recommendations. Additionally, the review highlighted the potential for misleading user communication within these apps, further exacerbating the risks associated with their use. Overall, the review underscored the need for rigorous validation and regulatory oversight to ensure the safety and effectiveness of AI dermatology apps.

Transparency and Regulatory Oversight: A Crucial Gap

The review also shed light on the concerning lack of transparency observed in the development of these apps and the functioning of their algorithms. Many AI dermatology apps do not provide clear insight into how their algorithms operate or the methodologies used in their development. In fact, 58.5% (24 apps) did not give any information on data used to train the AI model, and the ones that did, only offered vague and general descriptions. Furthermore, 51.2% of apps did not carry any information about the algorithm.

Moreover, the absence of FDA approval or other regulatory marks is a significant issue. None of the apps were approved by the FDA, and only 4 out of 41 apps had disclaimers about this. The lack of regulatory oversight means that these apps may not undergo rigorous evaluation processes to ensure their safety, efficacy, and compliance with medical standards. Consequently, users may be exposed to potential risks associated with inaccurate diagnoses, inappropriate treatment recommendations, and breaches of privacy.

The Critical Need for Clinician Involvement

Dermatologists possess invaluable expertise in diagnosing and treating skin conditions, making their input indispensable in app development. However, the findings revealed minimal clinician involvement in the creation of these apps. Only 39% of apps included dermatologists' input. This raises concerns about the ability of the apps to provide accurate diagnoses and appropriate treatment recommendations. Collaborating with dermatologists can enhance the clinical validity and utility of AI dermatology apps, ultimately improving patient outcomes and promoting trust in these digital healthcare solutions.

Privacy and Data Usage: User Concerns

User privacy concerns loom large over AI dermatology apps, particularly regarding how these apps store and utilize sensitive user data that is personal and potentially stigmatizing. However, the review revealed a lack of transparency regarding data storage and usage practices among many apps. Only 29.3% apps declared not to store any of the submitted images, and out of the 16 apps that did store the images, only 12 reported using secure cloud servers.

Upholding medical confidentiality and ethical standards is of utmost importance when it comes to medical apps because if this data is mishandled, it could lead to breaches of privacy, identity theft, or even discrimination. Therefore, stringent data protection measures are essential to maintain user trust and compliance with legal and ethical obligations in healthcare settings.

Evaluating Efficacy: Where’s the Evidence?

Even though most apps claim various benefits including diagnostic capabilities, supporting evidence to prove those claims is scarce. In fact, only 5 apps out of 41 were backed by evidence from peer-reviewed journal publications about their efficacy and reliability. Therefore, users should approach health apps with a critical eye, looking for independent studies and peer-reviewed research supporting the claims made by the app, and their adherence to established medical guidelines. By scrutinizing the evidence base of health apps, you can make more informed decisions about their suitability and reliability for managing your health concerns.

Future of AI in dermatology apps

Moving forward, enhancing AI dermatology apps requires several crucial steps. Firstly, standardization of validation processes and guidelines are needed to ensure consistent quality and reliability across apps. Additionally, improved regulation is needed to make sure the apps meet strict standards for safety, effectiveness, and privacy protection. Continued investment in research is vital to advance the capabilities of these apps, including refining AI algorithms and expanding the evidence base supporting their use. Collaboration between developers, healthcare professionals, and regulatory bodies is the key to driving innovation while safeguarding user interests.

By implementing these measures, AI dermatology apps can evolve into trusted and effective tools, offering valuable support in skincare management while upholding the highest standards of safety and quality.

Conclusion

In conclusion, the scoping review offers a cautiously optimistic outlook on the utilization of AI in dermatology. While AI dermatology apps hold promise for improving access to care and patient outcomes, their current state highlights the need for careful consideration of validation, regulation, and clinician involvement to maximize benefits while minimizing risks.

References:

1. Hogarty DT, Mackey DA, Hewitt AW. Current state and future prospects of artificial intelligence in ophthalmology: A review. Clin Exp Ophthalmol. 2019;47:128–39. DOI: 10.1111/ceo.13381
2. Chartrand G, Cheng PM, Vorontsov E, Drozdzal M, Turcotte S, Pal CJ, et al. Deep learning: A primer for radiologists. Radiographics. 2017;37:2113–31. https://doi.org/10.1148/rg.2017170077
3. Nasr-Esfahani E, Samavi S, Karimi N, Soroushmehr SMR, Jafari MH, Ward K, et al. Melanoma detection by analysis of clinical images using convolutional neural network. Conf Proc IEEE Eng Med Biol Soc. 2016;2016:1373–6. DOI: 10.1109/EMBC.2016.7590963
4. De A, Sarda A, Gupta S, Das S. Use of artificial intelligence in dermatology. Indian Journal of Dermatology. 2020;65(5):352. doi:https://doi.org/10.4103/ijd.ijd_418_20
5. Wongvibulsin S, Yan MJ, Pahalyants V, Murphy W, Daneshjou R, Rotemberg V. Current State of Dermatology Mobile Applications With Artificial Intelligence Features: A Scoping Review. JAMA Dermatology. Published online March 7, 2024. doi:https://doi.org/10.1001/jamadermatol.2024.0468