The Emergence of Cutaneous Digital Twins in Dermatology: A Systematic Literature Review

Digital twins (DTs) have made a significant impact on many industrial sectors, including healthcare and biomedicine, where they are being applied to patient-specific digital twinning, personalised medicine, predictive diagnostics, and surgical planning. While the literature is sparse, studies are emerging, with a particular focus on dermatology because many skin-related issues are inherently multimodal, longitudinal, and patient-specific. This has led to the development of cutaneous digital twins (CDTs), which establish an automated, live, and bidirectional relationship between the physical cutaneous system and its digital counterpart. A CDT is therefore a patient-specific and continuously updated virtual model of the skin. To characterise the current technical landscape of this emerging field, a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-informed systematic literature review was undertaken. This identified 426 records; following removal of duplicates and non-article records, 248 underwent title and abstract screening, resulting in 16 retained studies. Three domains were identified from the retained studies, with physics-based CDTs for transdermal drug delivery and percutaneous absorption being the most developed, followed by biomechanical CDTs for skin modelling, growth, and tissue expansion, and disease-specific and AI-enabled CDTs being the least mature. Cross-cutting gaps in CDT development were identified, including limited multimodal integration, underdeveloped longitudinal updating, inconsistent and preclinical validation, weak clinical workflow integration, and the overarching gap between technical enablement and CDT realisation. Overcoming these gaps would require a shift toward workflow-oriented architectures. Manufacturing DT principles can offer a roadmap for this transition by providing an architectural model for connecting simulation, data, feedback, and optimisation within a governed lifecycle. For clinical translation, a human-in-the-loop layer should be introduced within the architecture to ensure that CDT outputs are interpreted appropriately and iteratively refined by clinicians for effective diagnosis.