When it comes to PK/PD modeling, many researchers stick with the tools they know, even when those tools may not be the most efficient or insightful. What if you could achieve better results in less time with software that combines cutting-edge algorithms with a streamlined, user-friendly interface? Monolix offers a fresh approach, yet misconceptions often hold researchers back. This article addresses the most often asked questions about Monolix, giving clear answers to help you decide if it’s the right choice for your model-informed drug development needs. 

Can Monolix Handle All Modeling Phases, from Exploration to Regulatory Submission? 

You can use MonolixSuite applications throughout the entire modeling process, from exploratory up to the submission phase. Analysis obtained with Monolix is accepted by regulatory agencies such as the EMA and FDA. FDA and EMA hold Monolix licenses, and in the software documentation [link] we specify the necessary files for submission, ensuring quality control and consistency. 

“The Monolix Suite is a cornerstone of our toolkit. Three key criteria make it indispensable for us: Reliability of integrated tools in complex modeling on tight schedules; Industry-level quality and adherence to rigorous industry standards; Swift and comprehensive customer support for technical and scientific queries.” – Matthias Machacek, Managing Director, LYO-X 

Why Does Monolix Rely Only on SAEM, and is it Enough for Accurate Estimation? 

SAEM is designed to work well for both simple and complex models, as well as for continuous and non-continuous data. The FOCEI algorithm, widely used in NONMEM, relies on a first-order approximation, and can lead to biased estimates for sparse or very variable data, large complex models, or joint models. SAEM’s stochastic nature overcomes these limitations.  

Monolix has fine-tuned SAEM’s implementation to enhance performance. Most of the time, you just click the run button to start the estimation process without having to change any setting. Auto-initialization of parameters, their graphical visualization, and simulated annealing to avoid local minima – all improve the optimization process. And for especially complex situations, the algorithm settings can be adjusted directly in the GUI. 

Are Monolix Results Truly Reproducible, Even with Stochastic Algorithms? 

When SAEM algorithm runs with the same initial parameter estimates and the same random seed – both information saved in the run file–, as well as the same operating system, Monolix always produces the same result.  

Start of the estimation from a different initial point or with a different seed, due to the finite number of iterations of the algorithm, might give a slightly different solution. Tools like the convergence assessment (multi-start from different initial values and different seed) in Monolix help to run sensitivity analysis to check if the observed difference is statistically insignificant.  

Can You Trust Monolix to Identify Model Instability? 

Monolix provides clear warnings when convergence criteria are not met or when errors arise. 

If the estimation process fails, for example when the model gives Nans, users receive a clear error message about it. When convergence criteria are not met within the maximal number of iterations, then Monolix provides a warning message, parameter estimates and a dedicated diagnostic that track parameter trajectories to help find potential issues with convergence. 

Instead of simply failing, Monolix offers insight into where adjustments might be needed, giving users a chance to improve their models. 

Can You Use NONMEM Datasets Directly in Monolix without Complications? 

Switching between software can raise concerns about dataset compatibility, but Monolix supports datasets like those used in NONMEM. While there are some differences, a NONMEM-type dataset can be used directly in Monolix with minimal adjustments. Those usually can be done with the built-in data formatting module.  

Monolix uses model independent dataset format which allows greater flexibility. For example, the same dataset with information only about dose amount and administration time can be used with a single absorption or a double absorption model. There is no need to manually duplicate data rows, as is sometimes needed in NONMEM. 

How Does Monolix Ensure Transparency in its Models and Equations? 

Each model from the Monolix’s library is accessible for verification using a built-in editor. Some models in Monolix use macros (like pkmodel for standard compartmental models) that correspond to standard equations and comparable to methods in NONMEM (e.g., ADVAN 1-4 routines). ODE-based models use typical mathematical formulas and users can write their own equations.  

How Does Monolix Combine Ease of Use with Flexibility for Time-Varying Covariates? 

Monolix’s graphical interface handles constant covariates, reducing the need for manual coding and improving model efficiency – Monolix knows in advance which covariate effect parameters are related to which fixed effects. Time-varying covariates, or non-standard parameter-covariate relationships, are included directly in the structural model, which gives the flexibility to handle advanced modeling requirements. 

“I have been using Monolix for 9 years now and it has become my main tool for pharmacometrics activities. Several reasons for that: the GUI is intuitive, the documentation is clear and comprehensive, the plot generation is straightforward, and the interactions between the different tools of the Suite are handy.  

This makes the software efficient to use daily for various types of data and analyses and fastens the learning curve for new users and trainees. In addition, support has proven to be reactive and contribute to solve issues. Finally, the continuous improvement of the Suite over the years took modeler feedback into account thus making it even more user-friendly. I’m looking forward for future evolutions in the upcoming years. “- Vincent Madelain, Pharmacometrician, SERVIER 

How Does Monolix Simplify Likelihood Handling Compared to Manual Approaches? 

Monolix simplifies analysis by automatically deriving likelihood functions from the model definition for censored data with M3 or M4 method, non-continuous data, or joint models. This contrasts with NONMEM, where users must write these formulas manually – a process that can be time-consuming and prone to error. Automatic handling of likelihoods streamlines modeling without compromising accuracy. 

What Options Does Monolix Offer for High-Performance Computing? 

The core of Monolix is written in C++, a high-performance programming language, ensuring fast and optimized computations. Monolix supports automatic parallelization by distributing calculations across multiple cores on a single machine.  

For users requiring more extensive computational resources, Monolix supports distributed parallel computing through the MPI license, which handle distributed calculations across multiple machines (e.g cluster). 

Monolix comes with lixoftConnectors, an R-API, for efficient scripting in R of workflows and management of large-scale simulations. Each click in the GUI can be coded using a function in an R script, while calculations stay in MonolixSuite C++ engine. This capability is particularly valuable for automating repetitive tasks, reproducibility or integrating Monolix into a larger computational pipeline. 

Further Advantages 

MonolixSuite offers a graphical interface, out-of-the-box plots and built-in model libraries to streamline the modeling process. It helps users to focus on scientific analysis rather than technical implementation.  

Just the fact of being able to zoom in with a click or to highlight the same individual across plots makes it much easier to understand if your model is good or not and improve it. 

“The MonolixSuite is an absolute game changer for pharmacometrics work. The ability to seamlessly move from population PK modeling in Monolix, to simulation in Simulx, and then exposure calculation in PKanalix expedites my workflow. MonolixSuite makes my work more efficient and allows me to get results to clients faster. “- Jessica K. Roberts, Senior Director, Pharmacometrics, Allucent 

Sampling from the conditional distribution gives richer diagnostic insights from plots, especially when shrinkage is high. It also allows researchers to run unbiased statistical tests, such as Pearson correlation tests for covariates or t-test for correlations between random effects, that help in covariate analysis. In addition, it is used in the novel automatic model building methods COSSAC, [link], which needs much less iterations and opens door to covariate search for complex models with long runs.  

Interconnection between all MonolixSuite applications, such as Simulx for simulations and PKanalix for NCA, enables seamless workflows, reducing the need for users to switch between multiple tools. With one click your data, model, and estimates from a Monolix project can be transferred to Simulx to build a new simulation. This unique environment simplifies complex projects and improves collaboration.  

What’s next?

As pharmacometrics evolves, Monolix continues to set the standard for modern modeling tools. Through this article, we’ve addressed the most frequent questions and concerns, highlighting how Monolix combines cutting-edge capabilities with an intuitive design to meet the diverse needs of researchers. 

Trusted by leading companies and institutions worldwide, Monolix is constantly evolving with user-driven improvements. By choosing Monolix, you’re not just selecting software—you’re joining a forward-thinking community dedicated to advancing pharmacometrics science with confidence and success. 

If you’d like to learn more about Monolix, schedule a live demo.