Every projection rests on a chain of evidence: the twin is calibrated to published phase-3 endpoints, then held to external echocardiography, cohort scatter and parameter-sensitivity checks. The figures below are reproduced natively from the validation analyses.
Solomon SD et al., JACC 2024. The twin recovers the left-atrial volume endpoint on target, carries the placebo-corrected slope within published bounds, and reproduces the echocardiographic effect profile.
| Endpoint | Sim | Target |
|---|---|---|
| LA ETD (mL) | −6.13 | −6.13 |
| LA slope (mL/kg) | +0.52 | +0.55 |
| E/e′ ETD | −1.97 | −0.79 |
LA volume regression is driven jointly by weight loss (plasma-volume reduction) and E/e′ improvement (reduced LA stretch). No new ODE state required.
Each agent's weight-loss parameter (e_W) is back-solved to its published headline; the same coupled ODE then projects every downstream state. Projected 52-week trajectories in obesity-related HFpEF.
Per-patient changes from a simulated cohort overlaid on the published SUMMIT scatter, by diabetes status. The simulated correlation matches the published value at the cohort level.
Each parameter perturbed ±20% from baseline; bars show the resulting change in each endpoint. The drug-effect parameters dominate; downstream coupling terms are second-order.