Answers are grounded in this paper and cite their sources.
Digital twins (DTs) use simulation models to continually inform real-world decision making, and real-world data is used to continually improve DTs' simulation models. This bidirectional feedback between a DT's simulation models and the corresponding real-world setting is a defining feature of DTs. However, in rapidly evolving fields such as digital health, the continuity of this bidirectional relationship can be challenged by the speed of real-world progress. Advancements in science and technology can be so rapid that the simulation of upcoming decisions can repeatedly require new input or output variables that prior data do not include (e.g., due to novel sensors or interventions). This results in mismatches that we term data impoverishment: the absence of variables from prior data needed for a DT's models to simulate upcoming real-world decisions. While recent literature implicitly recognizes data impoverishment in brief, the challenge is yet to be examined in detail. In this perspective, we formally define data impoverishment and provide recent examples of data impoverishment to highlight the immediacy of this challenge in digital health. We also outline key factors that determine how often a DT will be affected by data impoverishment. Finally, we outline approaches to plan for and address data impoverishment, provide examples of how it has been addressed in prior work, and conclude with a discussion of future research to better address data impoverishment in rapidly evolving fields such as digital health.
No takes yet. Share an insight, caveat, or question.
Gazi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: