CardioMetaIQ projects one patient's cardiometabolic trajectory under GLP-1 receptor agonists — a nine-state coupled ODE calibrated against published phase-3 evidence. Every projected number is sourced, hedged and auditable.
CardioMetaIQ is a mechanistic digital twin of the obese HFpEF heart. A clinician enters one patient's phenotype; the twin integrates a coupled system of differential equations and returns a graded 52-week response prediction for GLP-1 receptor agonists — with the full trajectory, validation and uncertainty behind every number. Not a black box, and nothing is auto-prescribed.
Three deterministic steps. The same engine that reproduces trial-level endpoints in silico projects them forward for a single patient.
Baseline weight, LVEF, E/e′, NYHA class, NT-proBNP, diabetes status, GDMT and the candidate agent & dose — the entered identity that scopes every downstream claim.
A nine-state coupled ODE, forced by a GLP-1 drug fraction f(t), is integrated with RK4 at half-week steps across 52 weeks against the calibrated trial anchors.
A 0–100 response score, placebo-corrected effect sizes (ETD), full state trajectories and a Monte-Carlo uncertainty band — each hedged and traceable to source.
Before it predicts, the twin must recover what the trials already measured. Against the STEP-HFpEF echocardiography substudy it lands the left-atrial volume endpoint on target — and carries the placebo-corrected slope within published bounds.
See the full calibration figures →A guided motion demo of the clinical workspace — phenotype entry, the graded response, safety screen, 52-week trajectories and the validation tabs — narrated as a single patient passes through.
Watch the demo →