Sage Bionetworks wins ARPA-H contract to test ICU digital twins

Sage Bionetworks has won an ARPA-H contract to build Twin-TECC, a platform that will evaluate ICU digital twins from the CIRCLE program against FDA guidance and help prepare them for regulatory review. The project gives research teams a shared testing environment while aiming to separate promising models from ones that are not ready for clinical use.

Why it matters: - Twin-TECC is designed to answer a key question before AI digital twins reach the ICU: whether a clinician can trust them to inform care. - The platform could speed up regulatory readiness for patient-specific models built to predict how critically ill patients respond to interventions. - A shared evaluation system could make it easier to compare models across teams instead of leaving each group to grade its own work.

What happened: - Sage Bionetworks received a contract award from the Advanced Research Projects Agency for Health, or ARPA-H, to build Twin-TECC, which stands for Testing, Evaluation, Clinical alignment, and Compliance of Digital Twin Models. - The platform will evaluate digital twins developed through ARPA-H's Critical Illness Immunological Reprogramming and Control Point Learning Engine, or CIRCLE, program. - ARPA-H is awarding Sage Bionetworks up to $18.3 million over five years to serve as CIRCLE's independent evaluation partner. - The announcement was made in Seattle on Sept. 23, 2026.

The details: - Digital twins are computational models of individual patients that predict responses to interventions. - CIRCLE teams are building real-time datasets on critical illness progression, patient-specific immune system models, and methods to act on those predictions in the ICU. - Sage's independent evaluation work will assess each CIRCLE model against FDA guidance and established validation standards. - The evaluation will use stratified patient data that model developers do not see. - Twin-TECC will also give CIRCLE teams a shared sandbox to bring in models, run simulations with patient data, simulate clinical trials, and generate documentation regulators are expected to review. - The platform accepts models in the formats the field already uses. - Model developers control access, which is meant to protect intellectual property while allowing rigorous evaluation. - Sage says Synapse, one of the NIH-listed generalist repositories, manages more than 4.5 petabytes of biomedical research data. - Sage says Twin-TECC builds on 15 years of experience with governed biomedical data infrastructure, more than 20 DREAM Challenges, and tools created for the ARPA-H Biomedical Data Fabric Toolbox. - Sage is building Twin-TECC with subawardees InSilicoTrials Technologies, NYU Grossman School of Medicine, and The Jackson Laboratory. - Modeling teams and clinical partners can contact the Sage Bionetworks team at CIRCLE_AP2@sagebionetworks.org. - Sage Bionetworks is a Seattle-based nonprofit research organization founded in 2009. - Sage develops and operates Synapse and focuses on open science and governed data commons for sensitive biomedical data. - More information is available in Sage Bionetworks' announcement.

Between the lines: - The project reflects a broader shift from building AI models to proving they are trustworthy, comparable, and ready for oversight. - The combination of an independent evaluator and a shared sandbox suggests ARPA-H wants reproducible evidence, not just strong internal performance claims. - Regulatory alignment is a central theme here, which signals that clinical adoption will depend on validation standards as much as on model sophistication.

What's next: - Sage Bionetworks will build out Twin-TECC as the evaluation platform for CIRCLE models. - Research teams and clinical partners interested in the platform are expected to begin engaging with Sage for model testing and simulation. - The five-year award period suggests the evaluation framework will evolve alongside the CIRCLE program's digital twin models.

The bottom line: - ARPA-H is funding the infrastructure needed to test whether ICU digital twins are trustworthy enough for clinical review, not just innovative enough to build.