Industrial versus human digital twins: a comparative survey of concepts, architectures, enabling technologies, and open challenges

The digital-twin paradigm has scaled from engineering systems to smart cities, but extending it to living individuals introduces biological, psychological, and social complexity that conventional twins never had to model. The resulting field of human digital twins (HDTs) lacks a clear boundary with its industrial predecessor, and its literature shows inconsistent terminology and divergent architectures. This survey draws that boundary. Following a PRISMA-informed protocol, we analyse 57 studies retrieved from six databases and compare the industrial digital twin (DT) and the HDT across definition, data, architecture, enabling technologies, applications, and governance. Industrial DTs span a spectrum from monitoring-only digital shadows to fully integrated twins whose physics-based models support automated closed-loop control certifiable within defined tolerances. HDTs are distinguished less by the computational nature of their models than by their coupling to a specific living individual, which demands probabilistic multi-domain modelling, heterogeneous sensor fusion, predominantly advisory human-in-the-loop operation, and strict ethical and regulatory compliance. Significantly, a recent scoping review found that only 12.08% of healthcare HDT studies meet formal digital-twin criteria. A unified taxonomy extending the Model-Shadow-Twin paradigm to human systems, alongside comparative frameworks and a five-part research agenda, provides a coherent roadmap for this fragmented field.