Full laboratory Explore
Linked pressure–volume loops, traces, circulation map and measurements; every control, intervention and drug; a baseline and comparison snapshots.
Leaves out: A defined virtual patient, not a model of any individual.
What the simulator covers
One closed-loop model drives every view. 25 lessons (25 of 25 foundation topics), 30 cases, 9 conditions and 38 scenarios, with what each leaves out stated beside it.
What it is
A lumped-parameter model of the whole circulation: four heart chambers with time-varying elastance, systemic and pulmonary arteries and veins, valves that open and close with pressure, the pericardium and pleural pressure, and conserved blood volume. It is solved in your browser, and the pressure–volume loops, traces, circulation map and every number come from the same simulated beat.
What you can do, and where
Pressure–volume loops Explore › Heart mechanics
LV and RV loops, end-diastolic and end-systolic points, ESPVR/EDPVR lines, stroke work and arterial elastance, overlay or side-by-side comparison.
Leaves out: Ees from a single loop is not identifiable; the model's Ees is a parameter.
Waveforms Explore › Waveforms and timing
Right atrial, RV, PA, wedge, LA, LV and aortic waveforms by recording site, with labels, calipers, measured-signal artifacts and a hide-the-labels quiz.
Leaves out: No real recordings are included; the idealised and measured signals are generated.
Wiggers diagram and cardiac cycle Explore › Heart mechanics
ECG timing aid, pressures, volume, flows and valve states with a phase explanation at the cursor.
Leaves out: The ECG is a timing aid, not a modelled electrical signal.
Starling and venous return Explore › Circulation and volume
Cardiac function and venous return curves, mean systemic filling pressure, and manoeuvres such as passive leg raise and diuresis.
Leaves out: The venous return line is schematic; the function curve is generated by the model.
Valves and disease Explore › Valves and disease
Aortic, mitral and tricuspid lesions (acute and chronic), HOCM, tamponade, constriction, restriction, pulmonary hypertension and heart-failure phenotypes.
Leaves out: Pulmonary valve disease and combined lesions are not supported; constriction does not reproduce LV/RV discordance.
Shock and drugs Explore › Shock and drugs
Seven shock phenotypes and ten named drugs as qualitative multi-target mappings, with oxygen delivery from a Fick teaching model.
Leaves out: Drug strengths are qualitative, not doses; no organ-level outcomes or drug kinetics.
Heart–lung interaction, exercise and rhythm Explore › Heart–lung and provocation
Spontaneous and positive-pressure breathing, PPV and SVV with caveats; graded exercise, Valsalva and fluid challenge; PVCs, AF and AV dissociation.
Leaves out: No lung-volume effect on PVR; the exercise mapping is illustrative; dyssynchrony is not modelled.
Mechanical support and shunts Explore › Support and interventions
IABP timing, LV pump with suction, VA-ECMO, ECPELLA and ASD/VSD, with native-versus-device flow accounting.
Leaves out: Device curves are placeholders, not manufacturer data; no differential oxygenation; no RV support devices.
Structural interventions Explore › Support and interventions
Valve relief or repair, drainage and shunt closure, immediately after versus remodelled.
Leaves out: The remodelled state is an authored parameter set; no time course; AV fistula is not supported.
Cath lab Practice
Sequential recordings, artifact settings, reliability judgements before derived values, and output by Fick or thermodilution.
Leaves out: No station-by-station oximetry run for shunts.
Instructor tools More
Build and present sequences with locked controls and stepwise reveal; classroom projection; share links.
Leaves out: Sequences live in the browser and can be exported; share links hold state in the URL; no server-side snapshots.
What it does not do
- It does not simulate an individual patient and must not be used for patient care.
- No inertance, no septal geometry (ventricular interaction is a gain), no ischaemia-dependent contractility, no drug kinetics, no simulated remodelling.
- Teaching statements and answer keys are proposed content and have not been expert reviewed. See Sources & model details for how review and testing differ.
Technical details and development notes
Finding Hemo app 0.5.0 · content 0.5.0 · engine fh-closed-loop 0.4.0. Solver: RK4 at 0.2 ms, output at 1 kHz, run in a Web Worker; a state is called settled when three consecutive windows agree on eight summary measures. The scientific model document is docs/science/MODEL.md, the validation report docs/validation/REPORT.md and the milestone history docs/MILESTONES.md in the repository.
- Conservation: volume drift < 1e-9 relative over 100 beats with every circuit running; junction balance exact.
- Convergence and repeatability: core measures change < 1% at half the step; reruns are bit-identical; saved sessions continue the same trajectory; model migrations reproduce earlier results.
- Behaviour: directional tests for each intervention, phenotype, valve lesion, device and shunt; every case's intervene step is checked against the engine.