FracNeuPK — Documentation

FracNeuPK is a Fractional Neural Pharmacokinetics Simulator designed to simulate drug concentration-time profiles using a fractional neural differential equation approach, with user-defined covariates affecting drug elimination.

Key Features

• Fractional-order modeling of drug elimination (0 < α ≤ 1) to capture non-integer kinetics.
• Neural network correction term f_θ(t, C) for additional flexibility in dynamics.
• User-specified covariates affecting effective elimination rate (k_e,eff):
  • Age: Elderly patients typically have reduced clearance.
  • Renal function (eGFR): Adjusts clearance proportionally.
  • Lipophilicity: Lipophilic drugs tend to have slower apparent elimination.
  • Drug-drug interactions (DDI): Inhibitors reduce elimination rate.
• IV bolus dosing with user-defined dose and volume of distribution (Vd assumed 10 L for demo).
• Graphical output with Plotly for interactive concentration-time visualization.

Mathematical Model

The fractional neural DE solved is of the form:

D^αC(t) = f_θ(t, C) - k_e,eff·C(t)

Where:

• D^α is the Caputo-like fractional derivative of order α.
• f_θ(t, C) is a small neural network correction term.
• k_e,eff is the elimination rate adjusted for covariates.

User Inputs

• Drug Name: Any market drug name.
• Dose: Dose in mg.
• Fractional Order α: Between 0 and 1.
• Simulation Time: Total hours to simulate.
• Age: Patient age in years (optional).
• Renal Function (eGFR): mL/min (optional).
• Lipophilic Drug: Check if drug is lipophilic.
• Drug-Drug Interaction: Check if DDI present.

Outputs

• Interactive concentration-time plot.
• Table of time vs. concentration values (rounded to 2 decimals).
• Effective elimination rate used in simulation.

Notes

• This is a demonstration simulator; real pharmacokinetic parameters may vary.
• Volume of distribution (Vd) is assumed constant (10 L) for all drugs here.
• Fractional neural DE allows capturing non-classical kinetics not explained by standard one-compartment models.