Bridging TA1 (phaseless design engine) & TA3 (agentic operations)in a continuous self-learning loop
Qairnel (Paris & New York, spin-off of the Paris Brain Institute) is actively seeking clinical trial networks, academic prime contractors, and TA2 biostatistical partners to submit a joint proposal for ARPA-H SURPASS. We bring two production-tested technologies—CourseMap AI (TA1) and DocMemo (TA3)—already operating as a unified, closed-loop platform.
Operational Patient Digital Twins
Patented, Nature Comms–validated models (#1 vs. 56 methods), deployed operationally in <4 weeks post data-lock with 4 Top-10 Pharma sponsors.
Live “Virtual Waiting Room”
65,000+ screened individuals and 1,000+ AD trial referrals across 12+ sites, cutting in-clinic screen failure to <80% (vs. >90% industry standard).
Self-Learning Platform Loop
Longitudinal pre-screening trajectories directly calibrate TA1 digital twins, while trial outcomes continuously re-optimize TA3 patient allocation.
CourseMap AIFrom award-winning academic research to operational digital twins
For TA1, Qairnel brings CourseMap AI—a patented, award-winning spatiotemporal digital-twin engine validated in Nature Communications, where it ranked #1 against 56 competing methods in the international TADPOLE Challenge for predicting cognitive decline in Alzheimer’s disease.
Crucially, CourseMap AI is not merely an academic algorithm: it is an industrial-grade system utilized by 4 top-tier pharmaceutical sponsors, with proven statistical power gains on Sanofi’s MOVES-PD study and operational Phase 2 deployment delivering validated prognostic covariates in under 4 weeks from database lock.
Our vision for SURPASS TA1 is not to lock the consortium into a single proprietary model, but to architect an open simulation and digital-twin platform capable of hosting and combining multiple independent models. Empirical machine learning consistently demonstrates that ensemble aggregation of independent predictors yields superior accuracy and robustness.
Academic-to-Industrial Bridge: Born out of the Paris Brain Institute (ICM), our engineering team knows firsthand how to translate complex academic models into reproducible, regulatory-compliant (21 CFR Part 11 / FDA-aligned) operational software.
Award-winning, patented progression model validated in Nature Communications for predicting cognitive decline in Alzheimer’s disease.
Operationally deployed in an active Phase 2 trial—delivering validated prognostic covariates within 4 weeks of database lock.
Demonstrated significant variance reduction and power gain in Sanofi’s Phase 2 Parkinson’s disease study.
Industrial-grade software trusted and deployed across major biopharmaceutical clinical development pipelines.
Proven in real-world operations across 12+ partner clinical trial sites in the US and Europe.
Substantially increases in-clinic screening conversion, allowing sites to focus on patient visits rather than outreach.
Continuous community engagement of subjective memory complainers ready for Day-1 mobilization and dynamic multi-arm allocation.
Captures real-world cognitive risk profiles and trajectories during pre-screening to initialize TA1 digital twins.
DocMemoAn always-on “virtual waiting room” purpose-built for platform trials
For TA3, Qairnel brings DocMemo—a continuous digital engagement and qualification infrastructure that maintains a live “virtual waiting room” of phenotyped individuals with subjective cognitive complaints upstream of any specific protocol.
Ideal for Adaptive Platform Trials: Multi-arm perpetual trials require allocating the right sub-populations dynamically across evolving treatment arms over time. Maintaining an always-updated, pre-qualified cohort enables Day-1 enrollment without startup lag.
Tested at scale in early Alzheimer’s disease programs across 12+ partner sites with over 1,000 referred trial candidates, DocMemo has reduced screen failure rates to <80% (compared to >90% industry standard)—relieving clinical sites of recruitment friction so staff can focus on clinical visits.
A self-learning platform loopConnecting pre-screening operations to endpoint analysis
Qairnel is one of the very few organizations capable of connecting TA3 (pre-screening operations) and TA1 (digital-twin simulation) out of the box—creating a closed-loop learning system that continuously compounds trial efficiency, statistical power, and probability of success.
Longitudinal Risk Capture in the Waiting Room
DocMemo engages subjective memory complainers and collects longitudinal cognitive dynamics weeks or months before formal trial randomization.
Pre-Calibrated Baseline Digital Twins
Pre-screening trajectories combine with in-clinic biomarkers and imaging at randomization to initialize high-precision TA1 digital twins.
Real-Time Waiting Room Re-Optimization
Outcome data collected through randomization and trial follow-up continuously re-trains the TA3 screening models to identify and allocate the next wave of patients.
Continuous Trial Power & Adaptation
Prognostic covariates neutralize variance during the trial, reducing sample size requirements by 15%–25% and accelerating multi-arm adaptations.
Target indications & deployment readiness
Alzheimer’s Disease & Dementias
Pre-trained CourseMap AI manifolds + active DocMemo virtual waiting room (65,000+ screened individuals, 1,000+ trial referrals across 12+ sites).
Parkinson’s & Movement Disorders, Huntington’s, ALS
Proven Phase 2 covariate adjustment in Parkinson’s disease (MOVES-PD & Top-10 Pharma deployments) and calibrated longitudinal models in HD and ALS.
Mental Health & Neuropsychiatry
Leveraging DocMemo’s decentralized cognitive/behavioral capture and longitudinal state-space modeling to overcome high placebo response and heterogeneity.
Other Therapeutic Indications
Our multi-model TA1 simulation architecture and TA3 continuous pre-screening engine are disease-agnostic and can be tailored to high-unmet-need platform trials.
Partner with Qairnel on TA1, TA3, or integrated TA1+TA3
Contact our leadership team in New York and Paris to align on Solution Summary concepts (due November 30, 2026).