Diagnosing Blood-Pressure Changes from Model Trajectories

Reading a Pressure Trajectory as a Mechanistic Diagnosis

A simulated rise in blood pressure is an outcome, not a mechanism. The same pressure trajectory can arise because the heart pumps more blood per unit time, because systemic vessels offer more resistance to flow, or because renal sodium handling changes body fluid volume and thereby influences cardiac output and vascular state. The diagnostic task is to use the joint time courses of these quantities to identify which process is most consistent with the model’s behavior.

A single endpoint is often insufficient. Cardiac output and resistance may change together, while renal sodium excretion may initially fall and later rise as pressure and volume change. The order and persistence of these changes matter: an early deviation can be a candidate initiating event, while a later deviation may be compensation. This is a diagnosis of a simulated model’s mechanisms, not proof that the same mechanism caused a measured clinical pressure change.

Definition: In this tutorial, \(P_{\mathrm{art}}\) denotes modeled mean arterial pressure, \(CO\) denotes cardiac output, \(R_{\mathrm{sys}}\) denotes systemic vascular resistance, \(V_b\) denotes blood volume, \(E_{\mathrm{Na}}\) denotes renal sodium excretion rate, and \(I_{\mathrm{Na}}\) denotes sodium intake rate. A trajectory is the time course of one or more modeled variables after a defined initial condition or perturbation.

For a systemic circulation represented by a steady-flow approximation, pressure drop across the systemic circulation equals flow multiplied by resistance. If central venous pressure is included, the relationship is \(P_{\mathrm{art}}-P_{\mathrm{cv}}=CO\,R_{\mathrm{sys}}\), where \(P_{\mathrm{cv}}\) is central venous pressure. When central venous pressure is small relative to arterial pressure or is held constant, the commonly used approximation is:

P_{\mathrm{art}} \approx CO\,R_{\mathrm{sys}}.

The units must be compatible. For example, \(CO\) in L/min multiplied by \(R_{\mathrm{sys}}\) in mmHg·min/L gives pressure in mmHg. This relation is a useful model-level accounting identity or approximation; it does not by itself specify what caused either flow or resistance to change.

Separate the Pressure Components Before Naming a Cause

Compare a baseline time \(t_0\) with a later time \(t_1\). Under the approximation above, the ratio of pressures is the product of the ratios of cardiac output and resistance. Taking logarithms gives an exact decomposition of the modeled pressure ratio:

\ln\left(\frac{P_{\mathrm{art},1}}{P_{\mathrm{art},0}}\right) = \ln\left(\frac{CO_1}{CO_0}\right) + \ln\left(\frac{R_{\mathrm{sys},1}}{R_{\mathrm{sys},0}}\right).

This identity is useful because it quantifies the relative contributions to the endpoint change without assigning an order in which the components changed. It is not a causal decomposition: the model may couple \(CO\) and \(R_{\mathrm{sys}}\), and either may be downstream of another process. To diagnose a driver, inspect trajectories and, where possible, run a counterfactual simulation that removes or holds fixed a candidate mechanism.

A renal contribution requires an additional balance perspective. A lower excretion rate does not automatically mean that sodium is accumulating: intake may also be lower. The change in exchangeable body sodium depends on the difference between intake and losses. In a simplified model with renal excretion as the only loss route:

\frac{dN_{\mathrm{Na}}}{dt}=I_{\mathrm{Na}}-E_{\mathrm{Na}}, \qquad \Delta N_{\mathrm{Na}}(t_0,t_1) = \int_{t_0}^{t_1}\left[I_{\mathrm{Na}}(t)-E_{\mathrm{Na}}(t)\right]dt.

Here \(N_{\mathrm{Na}}\) is modeled body sodium content. If other sodium losses are represented, they must also be included in the balance. Sodium accumulation can promote extracellular-fluid expansion, but the mapping from sodium content to \(V_b\) depends on the model’s water-balance and compartment assumptions. Blood volume is therefore evidence about the pathway, not a substitute for checking the sodium balance.

Worked Example: Quantifying a Cardiac-Output Contribution

Suppose a hypothetical simulation has baseline \(CO_0=5.0\) L/min and \(R_{\mathrm{sys},0}=18.0\) mmHg·min/L. After a brief cardiac stimulation, the later values are \(CO_1=6.0\) L/min and \(R_{\mathrm{sys},1}=17.5\) mmHg·min/L. Assume central venous pressure is negligible or unchanged.

The baseline pressure predicted by the approximation is \(P_{\mathrm{art},0}=5.0\times18.0=90.0\) mmHg. The later pressure is \(P_{\mathrm{art},1}=6.0\times17.5=105\) mmHg, an increase of \(15.0\) mmHg.

The log contributions are \(\ln(6.0/5.0)=0.1823\) from cardiac output and \(\ln(17.5/18.0)=-0.02817\) from resistance. Their sum is \(0.1542\), which equals \(\ln(105/90)=0.1542\), to rounding. Cardiac output increased pressure, while the small resistance decrease opposed part of that increase.

This supports a cardiac-output-dominant component of the pressure rise. It does not establish that the cardiac change was the initiating cause: a separate model process could have raised output and pressure together. The intervention and timing provide further evidence about causal origin.

Use Timing and Balance to Evaluate Candidate Mechanisms

A trajectory-based diagnosis starts by aligning variables on the same time axis and identifying the perturbation time. Check that the baseline is stable, the reported values use consistent units, and the variables refer to the same model compartments and sampling times. Then ask which change appears first, which changes persist, and whether the observed sequence is compatible with the model’s equations.

Cardiac candidate: \(CO\) changes early and in the direction that accounts for much of the pressure change, while \(R_{\mathrm{sys}}\) is initially stable or changes later. A later resistance response may be compensatory.
Vascular candidate: \(R_{\mathrm{sys}}\) changes early and quantitatively explains much of the pressure change, while the early \(CO\) response is smaller or opposite in direction. Later changes in output or renal variables may be consequences or compensations.
Renal candidate: sodium balance becomes positive, followed by a sustained change in \(V_b\); pressure then changes through model pathways affecting flow, resistance, or both. This pattern is more informative than a brief excretion-rate change alone.

These are patterns to test, not universal diagnostic rules. A model may contain fast renal effects, slow cardiac adaptations, or direct couplings that alter the expected ordering. The model’s own structure determines which temporal sequence is plausible.

Worked Example: Distinguishing Sodium Retention From a Brief Excretion Dip

Consider a hypothetical model with constant sodium intake of 150 mmol/day. Before a perturbation, renal excretion is also 150 mmol/day. For the first two days after the perturbation, excretion is 130 mmol/day; on days 2 through 6, it is 155 mmol/day. Assume the model has no other sodium losses during this interval.

During the first two days, the balance is \(150-130=20\) mmol/day, so sodium accumulates by \(20\times2=40\) mmol. During the next four days, the balance is \(150-155=-5\) mmol/day, so the model loses \(5\times4=20\) mmol. The net change over six days is \(40-20=20\) mmol of sodium accumulation.

Suppose the same simulation shows \(V_b\) rising by 0.10 L during the first two days and remaining 0.10 L above baseline at day 6. The temporary excretion dip is not merely a rate fluctuation: given constant intake and the stated loss assumption, it creates a positive cumulative balance that is only partly reversed. The persistent volume change is consistent with a renal-volume contribution in this model. It would not, by itself, prove that sodium retention caused the pressure change; the pressure and \(CO\), \(R_{\mathrm{sys}}\) trajectories must also be checked.

Test the Diagnosis With Counterfactual Runs

Time ordering and component decomposition narrow the possibilities, but coupled systems can remain ambiguous. A counterfactual run helps: repeat the simulation with the same initial conditions and perturbation while selectively holding a candidate variable or mechanism at its baseline behavior. The difference in pressure trajectories estimates that mechanism’s contribution within the specified model and intervention. Such a comparison is meaningful only if the model remains well-defined under the intervention and the held variable is not inadvertently changed through another route.

Establish a reference.
Record baseline \(P_{\mathrm{art}}\), \(CO\), \(R_{\mathrm{sys}}\), \(V_b\), intake, and excretion. Confirm that the baseline trajectory is stable or describe its ongoing trend.

Locate the first divergence.
Compare the perturbed run with the reference over time. Identify which state or rate first departs and whether the pressure change follows or precedes it.

Check arithmetic and balance.
Use \(P_{\mathrm{art}}\approx CO\,R_{\mathrm{sys}}\) to check the pressure components. Integrate intake minus excretion and other modeled losses before describing sodium retention.

Challenge the candidate.
Run a clearly defined counterfactual, such as fixing a vascular tone input or preventing a change in renal sodium handling, and compare its pressure trajectory with the original run.

State the scope of the conclusion.
Report the mechanism supported by the model trajectories and intervention, and identify any remaining ambiguity or model assumption that limits the diagnosis.

Worked Example: A Vascular-Led Rise With Later Renal Change

In a hypothetical simulation, baseline values are \(CO=5.0\) L/min and \(R_{\mathrm{sys}}=18.0\) mmHg·min/L, giving \(P_{\mathrm{art}}\approx90.0\) mmHg. After a vasoconstrictor input, resistance rises to 20.0 mmHg·min/L by hour 1, while output remains 5.0 L/min. At hour 6, output has fallen to 4.8 L/min. Renal sodium excretion is initially unchanged, then falls below intake over the next two days, and blood volume rises later.

At hour 1, the pressure estimate is \(5.0\times20.0=100\) mmHg, a 10.0 mmHg rise attributable arithmetically to the resistance increase at that time. At hour 6, the estimate is \(4.8\times20.0=96.0\) mmHg. The fall in output partly offsets the pressure effect of elevated resistance.

The ordering supports a vascular-led initial response: resistance changes before output, sodium balance, and volume. Later positive sodium balance and increased volume may help sustain the pressure change, depending on the model’s pathways. A counterfactual that holds the vasoconstrictor input at baseline can test whether the early pressure rise disappears. One should not call the later renal response the initiating cause merely because it is present at the final time point.

Diagnostics, Alternatives, and Limits

Several checks can reveal a weak or misleading interpretation. First, verify that the pressure identity is approximately satisfied using values at matched times. A mismatch may indicate that central venous pressure is important, the model includes other pressure terms, or quantities were sampled or converted inconsistently. Second, inspect whether the baseline itself drifts; a change from an unstable baseline is not necessarily an effect of the perturbation. Third, distinguish a rate from a cumulative amount: excretion below intake for one interval can be outweighed by later excretion above intake.

The main alternatives answer different questions. An endpoint comparison is simple but can obscure the order of events and compensatory responses. A component decomposition quantifies how output and resistance combine at a chosen time, but does not identify causal direction. A trajectory comparison adds temporal information, while a counterfactual intervention gives stronger within-model evidence about the role of a selected pathway. None removes uncertainty about whether the model structure adequately represents physiology.

Blood volume also should not be equated directly with arterial pressure. Volume may affect venous return and cardiac output, while vascular tone, distribution among fluid compartments, and feedback processes modify the response. Similarly, a higher sodium excretion rate is not necessarily evidence against renal involvement: it may be a response to increased pressure after earlier retention. Interpret each series together with intake, the cumulative balance, and the model’s stated mechanisms.

Common pitfall: “Renal excretion is low at the final observation, so renal sodium retention caused the entire pressure rise.” This reasoning skips intake, cumulative balance, timing, and alternative causes. The correction is to calculate net sodium balance over the relevant interval, inspect when volume and pressure begin to change, and test the renal pathway with a defined counterfactual when the model permits it.

Key takeaway: Diagnose a simulated pressure change by combining the pressure-flow-resistance relationship with time ordering and sodium-volume balance. Describe a mechanism as supported by the model only to the extent that its trajectory and counterfactual evidence distinguish it from competing explanations.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions.

A simulation changes from \(CO=4.0\) L/min and \(R_{\mathrm{sys}}=20.0\) mmHg·min/L to \(CO=4.4\) L/min and \(R_{\mathrm{sys}}=20.0\) mmHg·min/L. What are the approximate baseline and final pressures, and which pressure component changed?
With constant intake of 100 mmol/day, excretion is 90 mmol/day for three days and 110 mmol/day for one day. What is the net sodium change over four days, assuming no other losses?
Why is a rise in blood volume observed after a pressure increase not sufficient to show that renal retention initiated the pressure change?
What does a counterfactual run add beyond an endpoint decomposition, and what is one limitation of its interpretation?

Solutions. (1) Baseline pressure is \(4.0\times20.0=80.0\) mmHg; final pressure is \(4.4\times20.0=88.0\) mmHg. Output rose while resistance was unchanged, so the component calculation attributes the increase to higher \(CO\). (2) The first three days accumulate \((100-90)\times3=30\) mmol; the last day loses \(110-100=10\) mmol. Net accumulation is \(30-10=20\) mmol. (3) The pressure increase may have occurred first and altered renal excretion, making the later volume rise a response rather than an initiating event. Timing and the model’s sodium balance are needed. (4) A counterfactual tests how the pressure trajectory changes when a specified mechanism is prevented or fixed. Its conclusion is conditional on the model structure and the validity of that intervention; it does not establish a mechanism in real patients.