OBJECTIVES

Virtual cohorts and synthetic patients are gaining traction to address data scarcity, privacy and ethical constraints in drug development.

Quinten Health in partnership with INSERM/CONSTANCES, one of the largest general population cohorts in EU is developing a regulatory-grade, data-integrative, disease-specific approach to accelerate and de-risk drug development and evaluation using AI, machine-learning (ML) and predictive modeling.

The tool supports:

1) identifying patient profiles for clinical decision,

2) characterizing disease progression,

3) emulating standard of care (SoC) – treatment effect and exposure,

4) simulating alternative scenarios (e.g. population, timing),

5) modeling cost to identify drivers and optimize resources.

METHODS

The tool is based on the French CONSTANCES cohort (220,000+ adults), collecting health, socioeconomic, and biological data linked with insurance and hospital records.

It uses disease-specific models (eg COPD, obesity) and methods such as clustering for patient identification, Bayesian / survival models, neural networks for disease progression and SoC, Markov models for medico-economic analysis.

Generative adversarial networks GANs support tool development which produces HTAs-relevant outcomes, including quality of life, healthcare resource use, and direct medical costs.

RESULTS

Based on pharmaceutical sponsors’ priorities, initial target indications include asthma, COPD, and obesity. A feasibility assessment evaluated the potential to generate HTA-grade evidence using the CONSTANCES-SNDS dataset.

While CONSTANCES may underrepresent severe cases in some diseases (e.g. COPD), it offers long-term follow-up and access to valuable para-clinical variables.

Linkage with SNDS adds value by enabling reconstruction of care pathways, treatments, and outcomes over time.

CONCLUSIONS

The CONSTANCES-SNDS integration offers a unique dataset combining depth, breadth, and long follow-up, making it fit-for-purpose for early decision-making in drug development, access, and evaluation.

By aligning regulatory-grade data with HTA and payer expectations, it provides a credible, scalable foundation to simulate real-world impact, reduce uncertainty, and inform strategic choices from target population definition to market access planning.