Runs in your browser Report an error © 2026 Medecsta
PKecsta: NCE Pharmacokinetic Tool All research tools

PKecsta: NCE Pharmacokinetic Tool

Study data

Paste from Excel: first column time, then one column per animal (or one mean profile). A header row is optional. Leave a cell blank or write BLQ where there is no reportable value; such cells are left out, never counted as zero.

No analysis yet

Enter the dose, paste a concentration-time table and press Analyse. The simulated example shows a full analysis with a known answer.

Non-compartmental analysis

    Compartmental model screen

    Fitted by least squares on ln(C), ranked by AICc. A model passes only if repeated searches agree, no parameter sits on a search bound, every point is predicted within 2-fold, every RSE is below 50 % and no parameter correlation exceeds 0.95. Pick a row to use it in the graphs.

    The model curve, residuals and observed-vs-predicted plots are in Graphs.

    Absolute bioavailability

    F = (AUCoral / Doseoral) / (AUCIV / DoseIV). Analyse the IV arm and an oral arm one after the other and both sides fill in from the mean AUC0-inf; or type the values.

    Methods and references

    • Units. Time h, concentration ng/mL, dose mg/kg, so clearance is in L/h/kg and volumes in L/kg. After an oral dose every clearance and volume is an apparent value (CL/F, V/F).
    • AUC. Linear trapezoidal rule. An oral profile starts from (0, 0). For an IV bolus C0 is back-extrapolated log-linearly from the first two points, or taken as the first observation when those do not decline.
    • Terminal slope λz. Log-linear regression on the last 3, 4, 5... points; the window with the best adjusted R² is kept (more points on a tie within 0.0001). An oral window never includes Cmax or anything before it. The points used are reported so the fit can be judged.
    • Extrapolation. AUC0-inf = AUC0-t + Clast/λz, from the observed Clast. MRT = AUMC0-inf/AUC0-inf; CL = Dose/AUC0-inf; Vz = CL/λz; Vss = CL × MRT (IV only).
    • Model fitting. Closed-form 1- and 2-compartment equations. Least squares on ln(C) (proportional error); 4000 seeded random starting points, the best 24 polished by Nelder-Mead. RSE % is the standard error of ln(parameter) from the Jacobian. AIC, AICc and BIC use k = p + 1. A model that would leave fewer than 2 residual degrees of freedom is not fitted.
    • Flip-flop. For one compartment with first-order absorption, swapping ka and the elimination rate gives the same curve. The tool reports the solution with ka faster than elimination; oral data alone cannot tell which is true, so compare with the IV half-life. CL/F is the same either way.
    • Checked against. Analytic solutions (exponential decay, Bateman equation) and numerical integration of each model's differential equations: pk.test.mjs. Not yet compared with Phoenix WinNonlin output.

    Gabrielsson J, Weiner D. Pharmacokinetic and Pharmacodynamic Data Analysis: Concepts and Applications, 5th ed. Swedish Pharmaceutical Press; 2016.
    Gibaldi M, Perrier D. Pharmacokinetics, 2nd ed. Marcel Dekker; 1982.
    Certara. Phoenix WinNonlin User's Guide: NCA computational rules (λz range selection).

    Runs in your browser — nothing is uploaded In review — research reference, not a dosing authority. Check anything that matters and report anything that looks wrong.