Overview
Circulation Studio is literature-informed educational software using synthesized hemodynamic data. It is not validated for clinical decision making or patient care. A plausible response, passing test or cited mechanism does not establish patient-specific predictive accuracy.
Three kinds of content coexist: established definitions/equations; literature-informed directional relationships; and original numerical/graphical teaching assumptions. The latter include phenotype endpoints, interaction gains, timing, response bounds, error demonstrations and successful-treatment scenarios.
What automated checks establish
Tests check finite numerical results, pressure/flow identities, lesion-grade behavior, flow conservation, timing consistency, treatment exceptions, search/navigation and selected regressions. Source-catalog checks ensure new clinical links are recorded. These are software and internal-consistency checks.
The 13 September 2026 multi-valve/medication/ventilation audit exercised 238,260 configurations and 9,828 sampled LV/RV respiratory-loop checks. It covered a defined matrix, not every continuous input combination or every clinical presentation. The dated audit report describes that scope; later content should not imply a new clinical validation study.
Major idealizations
- Settled drug, fluid and procedure endpoints; no general pharmacokinetic, cumulative fluid, renal or recovery-time model.
- Selected adult disease phenotypes; no patient-specific anatomy, age, measured compliance, vascular impedance or body-size calibration across all modules.
- Deterministic rhythm patterns and chosen rescue responses; no stochastic arrhythmia-risk, conduction-network or electrophysiology solver.
- Synthetic pressure contours and PV/echo volume; no measured conductance volume, Doppler, coronary flow or myocardial energetics.
- Single-compartment ventilation, fixed transmission/gas assumptions and bounded oxygen admixture; no regional lung or comprehensive acid–base model.
- Illustrative support flow and shunt geometry; no complete transient circuit accumulation, cannula streaming, hemolysis or regional oxygenator mixing.
- Reference/alert landmarks and success/error examples; no validated risk score, therapy recommendation or probability of adverse events.
How to use a surprising result
Compare baseline targets with final sampled values, confirm units and respiratory phase, identify native versus total/device flow, and inspect all selected pathologies and treatments. Check whether a number is measured from the trace, a formula estimate or a schematic proxy.
If the result contradicts flow balance, changes solely with viewport size, invents unavailable measurements or loses a lesion after another is added, that is a candidate software error. If it is an intentionally bounded extrapolation, it still needs its limitation stated. Documentation is not a reason to excuse a demonstrated bug.
Keeping the reference current
The model reference links every library module to a topic in a coverage inventory. Tests check that a newly added module has a documentation home, that article/section links resolve, and that section text and tables participate in search. Parameter tables reuse exported definitions where practical.
Coverage means the implementation has a documented home, not that every coefficient has been clinically validated. Changes to model behavior require an update to the relevant article, assumptions and references as well as numerical regression checks.
References & further reading