Presentation (CTI)

OBJECTIVES: AI-enabled digital twins may generate synthetic or in silico comparator evidence where conventional comparative data are difficult to obtain, particularly in rare diseases. Regulators recognise them as an emerging drug-development use case, but their relevance to HTA remains unclear. This research assesses current HTA use and explores stakeholder acceptability, barriers and requirements.
METHODS: A targeted review searched PubMed, Google Scholar and Value in Health (2015 onwards) for literature on digital twins, virtual patients, in silico trials and adjacent methods relevant to HTA, reimbursement, regulatory evidence generation or uncertainty. Grey literature searches covered European HTA bodies and regulators. Additionally, a structured online survey captured views from former European payer/HTA stakeholders on use cases, requirements, and barriers associated with HTA acceptance of AI-generated comparative evidence.
RESULTS: No HTA submissions using digital twin comparator evidence were identified. Available guidance is limited and cautious. NICE was the only HTA body identified as explicitly exploring appraisal methods for AI-generated data. A TLV post-launch economic follow-up described a digital twin approach for daratumumab real-world data; but not as formal reimbursement appraisal. Stakeholder insight was heterogeneous, with potential in single-arm trials, rare diseases, where conventional comparators are infeasible. The expected decision impact is supportive: contextualising treatment effects and uncertainty rather than replacing standard comparative evidence. Key requirements included transparency, external validation, patient-level data, uncertainty analysis, reproducibility and clinical plausibility review. Barriers included limited guidance, residual confounding, uncertainty quantification, generalisability and assessor trust.
CONCLUSIONS: Digital twin evidence use in HTA remains immature and exploratory. Stakeholder insight suggests AI-generated comparative evidence may be considered as supportive in selected evidence gaps, but payer acceptance remains conditional, heterogeneous and dependent on validation and transparency standards. Future research should define acceptability criteria and assess whether AI-enabled comparator evidence can support HTA decision-making by reducing uncertainty around comparative effectiveness.