Surgery 2026

Background: The National Academies of Sciences, Engineering, and Medicine (NASEM) define HDTs as virtual constructs that mimic a human system, are dynamically updated with data from its physical counterpart, predict trajectories, and inform clinical decisions. HDTs are often confused with predictive models or interactive simulations; their value depends on personalization, bidirectional updating, quantified uncertainty, and demonstrated clinical value. In anesthesia, the challenge is validating these capabilities before allowing increasingly autonomous clinical decisions.
Objective: To underscore the gap between predictive models, interactive simulations, and HDTs meeting NASEM criteria, and assess whether perioperative HDTs can support personalized anesthesia, improve patient-centered outcomes, and enable clinically safe autonomy.

Methods: We conducted a narrative review of recent publications, most published between 2022 and 2026, addressing HDT definitions, predictive models, patient simulation, closed-loop anesthesia, operating-room digital twins, artificial intelligence, and verification, validation, and uncertainty quantification (VVUQ). Studies were classified by individualization, dynamic updating, bidirectional interaction, autonomy, and clinical validation.

Results: Current systems support prediction, simulation, and supervised recommendations, but few demonstrate a continuously updated, bidirectional HDT. Closed-loop anesthesia improves control of physiological targets; evidence of improved patient-centered outcomes remains limited. Key issues include data integration, uncertainty, bias, model transparency, regulatory requirements, and responsibility for errors.

Conclusion: HDT development should follow a staged pathway from data integration to supervised and limited autonomy. Before algorithms adjust anesthetic infusions, external validation, prospective evidence, safety limits, quantified uncertainty, and continuous anesthesiologist oversight are required.

Updating soon...