Population States, Flows, and Cell-Count Accounting

Why Cell-Count Accounting Comes First

A cell-population model describes how the numbers of cells in defined groups change over time. Before choosing a growth law or estimating a rate, the model must answer a simpler question: which cells are counted in each state, and where does every flow go? If one differentiation flow is entered as an outflow from its source but omitted as an inflow to its destination, the model creates cells out of nothing or loses them without a stated mechanism.

Here, a compartment is a clearly specified cell group, and its state variable is the number of cells in that group. Compartments should be mutually exclusive at a given time: a cell belongs to one modeled compartment, not simultaneously to two. The equations below use cell-number flows, measured in cells per unit time. They describe a deterministic population approximation; counts are often treated as continuous in ODE models even though actual cell numbers are integers.

Definition: A cell-population state variable $X_i(t)$ is the number of cells in compartment $i$ at time $t$. An inflow adds cells to that compartment, an outflow removes cells from it, and a transfer flow removes cells from one modeled compartment and places those same cells in another. A net proliferation flux is defined here as the net number of cells added to the modeled population per unit time by proliferation.

The net-proliferation convention is important. A division event is not automatically equivalent to adding one cell: its contribution to a modeled compartment depends on how the parent and daughters are counted. In the equations here, the proliferation term is already expressed as the net addition of cells, so it can be added directly to a cell-count balance. If a model instead starts from division-event rates, it must first convert events into cell-number changes under its stated daughter-cell accounting rule.

The Compartment Balance Equation

For each compartment $i$, list all external inputs, net proliferation, transfer inflows, external losses, and transfer outflows. Let $U_i(t)$ be external influx, $B_i(t)$ net proliferation flux, and $L_i(t)$ external loss. Let $Q_{ij}(t)$ be the transfer flow from compartment $i$ to compartment $j$. Each of these terms has units of cells per unit time and is nonnegative under the stated flow definitions. Then the balance for compartment $i$ is:

\frac{dX_i}{dt} = U_i+B_i-L_i +\sum_{j\ne i}Q_{ji} -\sum_{j\ne i}Q_{ij}.

The first sum includes transfers into $i$; the second includes transfers out of $i$. This is an accounting identity, not a particular biological rate law. The modeler must supply expressions or data for the flows, but the signs and destinations follow from the bookkeeping. A term $Q_{ij}$ appears with a minus sign in the equation for $X_i$ and with a plus sign in the equation for $X_j$, because it is the same flow of cells.

Key result: When compartment equations are summed to obtain the total modeled cell count $X_{\mathrm{tot}}=\sum_i X_i$, every transfer between modeled compartments cancels. Thus, $$ \frac{dX_{\mathrm{tot}}}{dt} = \sum_i (U_i+B_i-L_i), $$ provided each internal transfer is included once as an outflow and once as an inflow. Internal differentiation redistributes cells; by itself, it does not change the total number of modeled cells.

Why Internal Transfers Cancel

The cancellation follows directly from the balance equations. Consider one transfer $Q_{ij}$, from $i$ to $j$. It contributes $-Q_{ij}$ to $dX_i/dt$ and $+Q_{ij}$ to $dX_j/dt$. In the sum over all compartment derivatives, these terms add to zero:

(-Q_{ij})+(+Q_{ij})=0.

Applying this to every internal transfer leaves only flows across the boundary of the modeled population: external influx, net cell production by proliferation, and external loss. This provides a useful audit. If a differentiation term changes the total count in a closed system with no proliferation or external loss, the equations contain an accounting error or the assumed compartments do not include all destinations.

Worked Example: One Compartment With Four Contributions

Suppose a hypothetical precursor-cell compartment contains $X(0)=800$ cells. At the current time, it receives an external influx of 120 cells per day, has a net proliferation flux of 30 cells per day, loses 50 cells per day, and differentiates 20 cells per day into a separately modeled compartment. Find its instantaneous rate of change.

The differentiation flow is an outflow from this compartment. The balance is therefore

\frac{dX}{dt}=U+B-L-Q =120+30-50-20.

Substituting the values gives

\frac{dX}{dt}=80\ \text{cells/day}.

At that instant, the precursor count is increasing at 80 cells per day. This is an instantaneous slope, not a claim that the count will increase by exactly 80 cells over every following day; that would require the flows to remain constant over the interval. The count $X(0)=800$ sets the current state but does not enter this balance separately because the flows have already been supplied as total cell-number rates.

Writing Equations for Differentiating Populations

For a multicompartment model, write one balance per state and use the same symbol and value for a transfer in both affected equations. In the next example, precursor cells $A$ differentiate into either $B$ or $C$. All flows are hypothetical total cell-number rates at the stated instant.

Worked Example: A Precursor With Two Differentiation Destinations

At a particular time, let $A=600$, $B=200$, and $C=100$ cells. The precursor receives 10 cells per hour from outside the modeled system and has a net proliferation flux of 18 cells per hour. It differentiates at 40 cells per hour into $B$ and 15 cells per hour into $C$. Compartment $B$ receives a further external influx of 5 cells per hour and loses 12 cells per hour externally. Compartment $C$ has an external loss of 8 cells per hour. There is no proliferation in $B$ or $C$. Write the three balances and calculate their current slopes.

For $A$, both differentiation flows are outflows:

\frac{dA}{dt}=10+18-40-15=-27\ \text{cells/hour}.

For $B$, the 40-cell-per-hour differentiation flow from $A$ is an inflow:

\frac{dB}{dt}=5+40-12=33\ \text{cells/hour}.

For $C$, the 15-cell-per-hour flow from $A$ is an inflow:

\frac{dC}{dt}=15-8=7\ \text{cells/hour}.

The total modeled count is $600+200+100=900$ cells. Summing the calculated slopes gives $-27+33+7=13$ cells per hour. The total external inputs are $10+5=15$ cells per hour, net proliferation adds 18 cells per hour, and external losses total $12+8=20$ cells per hour. Hence the total balance is $15+18-20=13$ cells per hour, in agreement. The internal differentiation flows cancel in the total.

Proliferation and the Meaning of a Flow

Proliferation requires a stated counting convention. In the balance equation, $B_i$ is a net cell-number contribution, not necessarily the number of mitoses per unit time. For example, if a modeled group replaces one counted parent cell with two counted daughter cells in the same group, one division event changes the count by $+1$, not $+2$: the parent is no longer counted after division. If daughters enter different compartments, the event must be allocated across those compartments and the parent removal included. The biological event description and the state definitions determine the correct terms.

This convention also prevents double counting. If a model defines proliferation as births of daughter cells per time while retaining the parent in its state equation, it must account for whether the parent is removed or remains counted. Adding two daughters without removing the parent in a replacement division would overstate the net increase. The safest practice is to write the event-level change in each state first, then express the result as a cell-number flux.

Worked Example: Converting Division Events to a Net Addition

A hypothetical compartment contains cells that divide at 12 events per day. Assume each event removes one counted parent from the compartment and places two counted daughters in the same compartment. There are no other flows. What is the net proliferation flux and the resulting balance?

Each event changes the compartment count by two daughter cells minus one parent cell:

\Delta X_{\text{per event}}=2-1=1\ \text{cell}.

At 12 events per day, the net addition is

B=12\ \frac{\text{events}}{\text{day}} \times 1\ \frac{\text{cell}}{\text{event}} =12\ \text{cells/day}.

Thus, $dX/dt=12$ cells per day at that instant. Simply multiplying 12 events per day by two daughters per event would give 24 cells per day, but that omits removal of the 12 parent cells and is not the net change under the stated replacement rule.

Diagnostics, Alternatives, and Limits

A reliable model can be checked at three levels. First, verify the sign of every term against the compartment definition: a transfer out must reduce the source count, and the same transfer must increase its destination. Second, check units: every term added in $dX_i/dt$ must have units of cells per time. A cell count cannot be added directly to a cell-number flow. Third, sum the equations and compare the result with the external boundary balance. Failure of the internal transfers to cancel is a strong signal of a missing or duplicated term.

Negative values of a state variable are also a warning. With nonnegative flows, a compartment should not be driven below zero in a physically meaningful implementation. A fixed outflow that continues when the compartment is empty can produce negative counts; in that case the flow specification is inappropriate near zero and must be reformulated to respect availability. Likewise, if a transfer leads to an unmodeled destination, it is an external loss relative to the chosen model boundary, not an internal transfer.

An alternative is to model a single aggregate cell count rather than separate compartments. That is sufficient when only total population is relevant and the internal composition does not affect the flows of interest. Separate states are needed when cell identity, differentiation destination, or compartment-specific processes matter. A tabular event ledger can be a useful intermediate representation: list each event, its source, destination, and cell-number change, then derive the ODEs. Regardless of representation, the same balance must hold.

These equations specify accounting, not the biological laws that determine how large each flow is. The flows may depend on states, time, or other modeled quantities, but their functional forms require evidence and a separate modeling decision. A correct balance cannot compensate for omitted compartments, poorly defined cell identities, or unsupported flow assumptions. It does, however, make those assumptions visible and provides a foundation for adding them consistently.

Common pitfall: “Differentiation removes 20 cells per day from the precursor, so the total population decreases by 20 cells per day.” This is wrong when those cells enter another compartment included in the model. The correction is to subtract the flow from the source and add the identical flow to the destination; the two contributions cancel in the total balance.

Check Your Understanding

Answer each question, then compare with the short solutions below.

A compartment has an external influx of 9 cells per hour, net proliferation of 4 cells per hour, external loss of 6 cells per hour, and transfer out of 3 cells per hour. What is its instantaneous slope?
Why must a transfer from compartment $A$ to compartment $B$ appear with opposite signs in the two compartment equations?
Two modeled compartments exchange 7 cells per day from $A$ to $B$. What is that exchange’s contribution to the derivative of their combined count?
A model reports a division-event rate in events per day. Is it automatically a net proliferation flux in cells per day? Explain.
In a model with three compartments, the sum of the compartment derivatives includes a negative differentiation term, but the destination compartment is included in the total. What should be checked first?

Solutions. (1) $9+4-6-3=4$ cells per hour. (2) The same cells leave $A$ and enter $B$, so the flow decreases one state and increases the other. (3) Its contributions are $-7+7=0$ cells per day. (4) No. The event’s net effect in cells depends on the number of cells entering and leaving each modeled state under the counting convention. (5) Check that the transfer is included as an equal positive inflow to its destination; internal transfers must cancel in the total.