Model Performance Metrics

📊 Institutional Performance Dataset (Public View)

MP Model (Vancomycin): 381 fits, generated 2026-08-26 14:32 UTC by marshall.
Two-Compartment Models: 371 fits, generated 2026-08-26 14:33 UTC.

This page displays key performance metrics from institutional pharmacokinetic modeling datasets.

Two-Compartment Model Performance

MP Model — Goti Goodness of Fit (371 patients, 494 fits)
% Converged
100.0%
Converged / Non
494 / 0
SS / level
5.756 (SD 8.639)
Pop pred MPE %
71.36 (SD 71.78)
Pop pred MAPE %
75.55 (SD 67.43)
Pop pred RMSE (mg/L)
8.582 (SD 5.648)
Pop pred NRMSE (%)
69.2 (SD 56.7)
1-Compartment (Matzke-Pai) goodness of fit — same patient cohort (N = 494 fits)

Identical metrics computed across all 1-compartment fittings for comparison with the Goti 2-compartment results above.

% Converged
100.0%
Converged / Non
494 / 0
SS / level
1.271 (SD 3.895)
Pop pred MPE %
24.01 (SD 47.82)
Pop pred MAPE %
38.97 (SD 36.77)
Pop pred RMSE (mg/L)
4.816 (SD 3.935)
Pop pred NRMSE (%)
36.2 (SD 31.4)
MP Model — Carreno Goodness of Fit (371 patients, 494 fits)
% Converged
100.0%
Converged / Non
494 / 0
SS / level
4.415 (SD 8.245)
Pop pred MPE %
87.42 (SD 75.61)
Pop pred MAPE %
89.92 (SD 72.66)
Pop pred RMSE (mg/L)
10.434 (SD 6.359)
Pop pred NRMSE (%)
82.4 (SD 61.2)
1-Compartment (Matzke-Pai) goodness of fit — same patient cohort (N = 494 fits)

Identical metrics computed across all 1-compartment fittings for comparison with the Carreno 2-compartment results above.

% Converged
100.0%
Converged / Non
494 / 0
SS / level
1.271 (SD 3.895)
Pop pred MPE %
24.01 (SD 47.82)
Pop pred MAPE %
38.97 (SD 36.77)
Pop pred RMSE (mg/L)
4.816 (SD 3.935)
Pop pred NRMSE (%)
36.2 (SD 31.4)

MP Model Performance — Vancomycin

MP Model — Patient Demographics — Vancomycin

N = 371    % Male = 53.9% (n = 371 with sex recorded)

Variable Mean SD Median Range N
Age (yr) 65.9 16.1 68.0 22.0 – 99.0 371
Height (in) 67.3 4.4 67.0 56.0 – 77.0 371
Weight (kg) 84.6 29.5 79.3 31.3 – 211.0 371
BSA (Meters Squared) 1.9 0.3 1.9 1.2 – 3.0 371
CrCl (mL/min) 66.3 35.5 59.6 7.9 – 216.2 371

Demographics from each patient's most recent Vancomycin fitting snapshot.

MP Model — Vancomycin — Creatinine Clearance Breakdown (mL/min/1.73 m²)

N = 371    Mean = 58.7 mL/min/1.73 m²  (SD 29.8, range 7.9–183.3)

Stage Category CrCl Range (mL/min/1.73 m²) N %
G1 Normal / High ≥ 90 51 13.7%
G2 Mildly Decreased 60–89 120 32.3%
G3a Mildly–Moderately Decreased 45–59 64 17.3%
G3b Moderately–Severely Decreased 30–44 69 18.6%
G4 Severely Decreased 15–29 57 15.4%
G5 Kidney Failure < 15 10 2.7%

Normalised CrCl = CrCl × 1.73 / BSA, using the fitting-time snapshot values.

MP Model — Vancomycin — BMI Breakdown

N = 371    Mean BMI = 28.7  (SD 8.7, range 11.8–68.6)

Category BMI Range N %
Underweight < 18.5 28 7.5%
Normal 18.5–24.9 103 27.8%
Overweight 25–29.9 107 28.8%
Obese 30–<40 99 26.7%
Morbid Obesity ≥ 40 34 9.2%

BMI computed from pt_weight / (pt_height × 0.0254)² using the fitting-time snapshot.

MP Model — Vancomycin — Fitted PK Parameter Group Distribution Vs Population Prior (all fittings)
Parameter Fitted (individual) Population
MeanSDMedianRangeN MeanSDRange
Vd (L) 52.693 21.370 48.279 8.310 – 160.630 494 55.343 19.526 20.350 – 137.150
Vd (L/kg) 0.614 0.102 0.631 0.238 – 1.176 494 0.650 0.000 0.650 – 0.650
Cl (L/hr) 3.9845 2.2728 3.3903 0.3801 – 12.5357 494 3.7270 2.0537 0.3146 – 12.0612
Ke (hr⁻¹) 0.0818 0.0470 0.0728 0.0091 – 0.2923 494 0.0679 0.0312 0.0088 – 0.2013
(hr) 12.15 9.05 9.52 2.37 – 76.31 494 12.99 7.85 3.40 – 78.70
Model Fit Bias — Fitted vs. Population Prior
Parameter MPE% MAPE% RMSE rRMSE% N
Vd (L) 9.2% 13.4% 8.83 24.4% 494
Vd (L/kg) 9.2% 13.4% 0.108 24.4% 494
Cl (L/hr) -3.1% 18.4% 0.9864 26.0% 494
Ke (hr⁻¹) -7.0% 26.0% 0.03496 34.3% 494
T½ (hr) 20.8% 33.4% 4.85 47.9% 494

MPE% (Mean Prediction Error): 1/n ∑[(Population − Fitted) / Fitted × 100%] — positive = population overpredicts individual fit; negative = population underpredicts; near zero = well-calibrated model. MAPE% (Mean Absolute Prediction Error): average magnitude of deviation regardless of direction — lower is better. RMSE (Root Mean Square Error): in the parameter's native units (L, L/hr, hr⁻¹, hr) — measures absolute spread of individual fits around the population prior. rRMSE% (Relative RMSE): RMSE normalised by the fitted value × 100% — scale-independent. MPE%: <20%: best   ≤30%: acceptable   >30%: caution — MAPE%: <30%: best   ≤50%: acceptable   >50%: caution — rRMSE%: <50%: best   ≤70%: acceptable   >70%: caution.

MP Model — Body-Size Diagnostics — Vancomycin (MP Model)

Per-fit individual (post-Bayesian fit) prediction error regressed against patient weight and BMI. N = 494 fits.

Fit (individual) prediction error stratified by BMI category
BMI category N fits Mean MPE % Mean MAPE % Mean RMSE (mg/L)
Underweight (<18.5) 41 2.34 3.84 0.565
Normal (18.5–24.9) 127 -0.24 3.51 0.630
Overweight (25.0–29.9) 143 0.49 2.85 0.444
Obese (30.0–39.9) 136 0.58 2.93 0.491
Morbid (≥40) 47 0.23 2.88 0.533
MP Model — Renal-Function Diagnostics — Vancomycin (Fit MPE % vs CrCl per 1.73 m²)

Per-fit individual (post-Bayesian fit) prediction error regressed against BSA-normalised CrCl (mL/min/1.73 m²). N = 494 fits.

Fit (individual) prediction error stratified by KDIGO CKD stage
CKD stage N fits Mean MPE % Mean MAPE % Mean RMSE (mg/L)
G1 — Normal (≥90) 66 0.23 3.11 0.534
G2 — Mildly Decreased (60–89) 159 0.71 2.50 0.371
G3a — Mild–Moderate (45–59) 90 0.64 3.07 0.480
G3b — Moderate–Severe (30–44) 93 0.48 3.20 0.521
G4 — Severely Decreased (15–29) 74 -0.82 4.00 0.767
G5 — Kidney Failure (<15) 12 4.70 6.15 1.327
MP Model — Bayesian Dosing Adjustment Accuracy — Predicted vs Achieved (Next Fitting)

Pairs a saved dosing recommendation (Predicted) with the fitted peak/trough/AUC₂₄ computed at the patient's next fitting (Achieved). Dose Match indicates both the recommended dose (within 1 mg) and dosing interval (within 0.1 hr) matched the active regimen at the next fitting.

Group Summary (dose-matched pairs only)
Parameter N Pairs MPE% MAPE% RMSE rRMSE% Mean Achieved (95% CI)
Peak (mg/L) 128 5.1% 12.6% 7.25 19.9% 35.91
(32.62 to 39.20)
Trough (mg/L) 128 4.7% 17.0% 5.85 27.1% 16.61
(13.53 to 19.70)
AUC₂₄ (mg·h/L) 128 4.2% 12.7% 145.8 18.7% 593.5
(516.8 to 670.2)