ADMET Predictor™

ADMET property prediction and
QSAR model-building application

Watch: ADMET Predictor Promo Video

Choose a Module:


What is the Simulation Module?

Unlike predictive the ADMET models described in other modules, the Simulation Module features a special set of predictive models that are not statistical in the sense of simply turning molecular structures into direct property estimations. In the Simulation Module the results are simulated using selected ADMET predictions, dose and predefined human physiology, and solving a deterministic, region-dependent system of differential equations.

This system is based on the “Human-Physiological-Fasted” version of the ACAT model of drug absorption employed in its entirety by our software GastroPlus™ reflecting compounds’ dissolution, permeation, transit and regional dependence along the intestinal tract. Users are advised to consult the GastroPlus documentation for further details inherent in the ACAT model.  The module predicts human intestinal absorption (HIA) for 1, 10, 100, 1000 mg dose.  It can also predict the oral dose to achieve a specific plasma concentration.  This module is currently available in ADMET Predictor 7.2 (it has not been ported to ADMET Predictor 8).  However, we anticipate that it will be available in the version after ADMET Predictor 8.


  • Beyond IC50 and simple PK models – Considerations for discovery chemists

    Robert Fraczkiewicz, Michael B. Bolger, Walter S. Woltosz

    What are the so-called “nonlinearities” in pharmacokinetics (PK) for multiple dosage of oral drugs? What are their causes and implications? While developing the Simulation Module [1] – an in silico tool for conducting PK simulations based on the Advanced Compartmental Absorption and Transit (ACAT™) model [2] – we have discovered a handful of interesting cases of nonlinear PK behavior.

    Learn More

SimHIA* Model

SimHIA* models simulate oral fraction absorbed in human. The only formulation variable changing here is dose. By default, all other parameters come from ADMET predictions, but can be optionally replaced by experimental values per compound. In fact, the default four SimHIA* columns reflect four different doses: 1 mg (SimHIA_1), 10 mg (SimHIA_10), 100 mg (SimHIA_100) and 1000 mg (SimHIA_1000).

SimDOSE Model

The SimDOSE (Simulated Optimal Oral Dose in Human) model in ADMET Predictor is used to assess oral dose levels, in mg, necessary to reach the desired unbound plasma concentration (Ceff) of drug candidates at steady state. The steady state is determined by the SimDOSE model as a state where administration of additional dose does not lead to further increase in unbound “peak and trough” plasma concentrations. The model accounts for possible limitation of fraction absorbed due to a low solubility or permeability of compound.

Most of the model inputs are by default calculated by ADMET Predictor models (or descriptors) from molecular structure, but others must be supplied by the user. The most important of these is the target effective concentration Ceff – the average unbound plasma concentration at steady-state. It must be supplied by the user in an input file, preferably in the form of per-compound auxiliary data. Another mandatory input is the total systemic clearance. Other inputs requiring the user’s attention are precipitation time, initial guess of the optimal dose, dosing interval and volume, particle radius, particle density, and body weight.