Now, what if we could predict whether a compound could reach an effective human dose very early in the development of a drug, or even incorporate that information into the design of the compound itself? That’s not as far-fetched as it may seem. In fact, several companies, including Roche¹ and J&J², are already doing just that. However, there are multiple approaches to human dose prediction, and each has its pros and cons.

Over the past decade, two major computational paradigms have emerged to address this challenge: purely data-driven, “AI” (really just advanced machine learning (ML) approaches and mechanistic physiologically based pharmacokinetic (PBPK) modeling. Increasingly, the industry is converging on hybrid approaches that combine the strengths of both.

Many ADMET properties are now routinely predicted using ML-based quantitative structure–property relationship (QSPR) models, often with excellent performance for endpoints such as solubility, permeability, plasma protein binding, or metabolic stability. Recent advances in deep learning and access to larger pharmaceutical datasets have further accelerated this trend. However, the prediction of integrated in vivo pharmacokinetic (PK) behavior, and ultimately human dose, has proven substantially more difficult. As Bassani et al. recently observed, despite the rapid increase in ML-based PK studies and larger datasets, “we did not observe a steady and substantial improvement in model predictivity.”³

The limitations are not surprising. Human dose is not a single intrinsic molecular property. Instead, it emerges from a complex interplay between absorption, distribution, metabolism, excretion, potency, formulation, route of administration, dosing frequency, and human physiology. Pure ML approaches attempt to infer these relationships directly from historical data, typically using molecular descriptors, fingerprints, or graph neural network representations as inputs. Several groups have demonstrated promising results using this strategy. Wang et al. developed ML models to predict intravenous PK parameters such as clearance and volume of distribution,⁴ while Lombardo et al. applied extensive datasets to improve prediction of human volume of distribution.⁵ Kosugi and Hosea compared ML-based clearance prediction against traditional bottom-up methods and demonstrated that ML methods can achieve competitive performance under certain conditions.⁶

These approaches offer clear advantages. ML models can be extremely fast, scalable, and well-suited for high-throughput screening environments where thousands of compounds may need to be evaluated rapidly. They are especially useful when the training domain is closely aligned with the chemistry space being explored. For early discovery teams, this enables rapid triaging of compounds before experimental data are available.

However, purely statistical approaches also have important limitations. Most notably, they often lack mechanistic interpretability. Predictions may be accurate within the training domain but difficult to extrapolate beyond it. ML models also struggle to explicitly incorporate dose-dependent phenomena, formulation effects, nonlinear PK, transporter-mediated disposition, or species translation. In many cases, they operate as “black boxes,” obscuring the physiological rationale behind predictions and making it challenging for scientists to understand why a compound is predicted to succeed or fail.

Mechanistic PBPK modeling provides a fundamentally different framework. Rather than directly learning dose from historical examples, PBPK models integrate compound-specific ADME parameters into mathematical representations of human physiology. Drug concentration-time profiles are simulated across tissues and organs using differential equations grounded in biology and anatomy. Importantly, these models can incorporate experimentally measured data, predicted properties, or both.

Historically, PBPK modeling was considered too computationally intensive for high-throughput discovery applications. Traditional PBPK workflows often required extensive manual parameterization and expert modelers. That perception has changed dramatically in recent years. Advances in automation, cloud computing, and streamlined PBPK architectures have enabled the emergence of high-throughput PBPK (HTPBPK) platforms capable of evaluating large virtual libraries rapidly and reproducibly.⁷

One of the most prominent examples is the High-Throughput Pharmacokinetics (HTPK) platform from Simulations Plus. HTPK combines ML-predicted or experimentally measured ADME properties with mechanistic PBPK simulation engines to rapidly estimate human PK and projected dose. Rather than replacing mechanistic modeling with AI, HTPK uses AI and predictive models to parameterize the mechanistic system. This distinction is critical. The PBPK framework provides physiological consistency, while ML models supply scalable predictions for the underlying inputs.

In practice, this hybrid approach offers several advantages over standalone ML PK prediction. First, it allows explicit incorporation of dose and route of administration, which are difficult for many QSPR models to capture directly. Second, the mechanistic structure improves interpretability: if predicted exposure is poor, scientists can determine whether the issue arises from clearance, permeability, solubility, first-pass metabolism, or another physiological process. Third, PBPK models naturally support species translation and scenario analysis, enabling simulations across preclinical species and humans using consistent mechanistic assumptions.

The performance of HTPK-style workflows has been encouraging. Naga et al. evaluated high-throughput PBPK predictions in discovery settings and demonstrated that mechanistic approaches could successfully inform early-stage compound prioritization and human PK estimation.¹ Importantly, these models were able to provide actionable guidance while maintaining the flexibility required for medicinal chemistry optimization.

The pharmaceutical industry has increasingly embraced these hybrid strategies. Roche’s SwiftPK platform is a notable example.¹ SwiftPK leverages Simulations Plus HTPK technology within Roche’s internal discovery workflows to enable rapid mechanistic PK predictions at scale. By integrating predictive ADME models with automated PBPK simulations, SwiftPK allows discovery scientists to estimate human exposure and projected dose liabilities much earlier in the design cycle. This type of platform exemplifies the broader industry movement toward combining AI-driven prediction with mechanistic modeling rather than viewing the two paradigms as competing alternatives.

Indeed, the future of predictive human dose estimation likely lies in the integration of ML and mechanistic approaches rather than exclusive reliance on either. ML models excel at rapidly predicting individual parameters from molecular structure and identifying complex nonlinear relationships in large datasets. PBPK models excel at integrating those parameters into physiologically meaningful simulations that can account for dose, route, formulation, and interspecies translation. Together, they form a complementary framework capable of supporting more informed and translational decision-making.

This convergence is particularly important as drug discovery increasingly prioritizes “developability” alongside potency. Medicinal chemists are no longer optimizing compounds solely for target affinity; they must also consider whether a molecule can realistically achieve therapeutic exposure at an acceptable human dose. Predictive dose estimation therefore becomes not just a DMPK exercise, but a central component of multi-parameter optimization.

Ultimately, the goal is not simply to predict PK endpoints more accurately, but to better understand how molecular properties translate into clinically viable medicines. Hybrid ML-mechanistic platforms such as HTPK and SwiftPK represent an important step toward that vision, enabling faster, more transparent, and more physiologically grounded predictions of human dose earlier than ever before.

(1)         Bassani, D.; Andrews-Morger, A.; Zhang, J.; Docci, L.; Cecere, G.; Pähler, A.; Belubbi, T.; Laye, P.; Shih, I.; Parrott, N. J. High-Throughput Physiologically Based Pharmacokinetic Model for Rodent Pharmacokinetics Prediction Using Machine Learning-Predicted Inputs and a Large In Vivo Pharmacokinetics Data Set. Mol. Pharmaceutics 2026, 23 (3), 1606–1617. https://doi.org/10.1021/acs.molpharmaceut.5c01317.

(2)         Van Rompaey, D.; Ray Chaudhuri, S.; Ahmad, M.; Cisar, J.; Van Den Bergh, A.; Ash, J.; Wu, Z.; Bryan, M. C.; Edwards, J. P.; DesJarlais, R.; Wegner, J. K.; Ceulemans, H.; Mitra, K.; Polidori, D. Toward Dose Prediction at Point of Design. J. Med. Chem. 2024, 67 (24), 22282–22290. https://doi.org/10.1021/acs.jmedchem.4c02385.

(3)         Bassani, D.; Parrott, N. J.; Manevski, N.; Zhang, J. D. Another String to Your Bow: Machine Learning Prediction of the Pharmacokinetic Properties of Small Molecules. Expert Opin Drug Discov 2024, 19 (6), 683–698. https://doi.org/10.1080/17460441.2024.2348157.

(4)         Wang, Y.; Liu, H.; Fan, Y.; Chen, X.; Yang, Y.; Zhu, L.; Zhao, J.; Chen, Y.; Zhang, Y. In Silico Prediction of Human Intravenous Pharmacokinetic Parameters with Improved Accuracy. J. Chem. Inf. Model. 2019, 59 (9), 3968–3980. https://doi.org/10.1021/acs.jcim.9b00300.

(5)         Lombardo, F.; Bentzien, J.; Berellini, G.; Muegge, I. In Silico Models of Human PK Parameters. Prediction of Volume of Distribution Using an Extensive Data Set and a Reduced Number of Parameters. JPharmSci 2021, 110 (1), 500–509. https://doi.org/10.1016/j.xphs.2020.08.023.

(6)         Kosugi, Y.; Hosea, N. Direct Comparison of Total Clearance Prediction: Computational Machine Learning Model versus Bottom-Up Approach Using In Vitro Assay. Mol. Pharmaceutics 2020, 17 (7), 2299–2309. https://doi.org/10.1021/acs.molpharmaceut.9b01294.

(7)         Naga, D.; Parrott, N.; Ecker, G. F.; Olivares-Morales, A. Evaluation of the Success of High-Throughput Physiologically Based Pharmacokinetic (HT-PBPK) Modeling Predictions to Inform Early Drug Discovery. Mol. Pharmaceutics 2022, 19 (7), 2203–2216. https://doi.org/10.1021/acs.molpharmaceut.2c00040.

At home, this may mean relying on AI to help us answer day-to-day questions. I’ve used it to plan summer vacations, help design our new interior paint colors, and help with the kids’ homework (yes, we all do it!).

At work, the gaps between users grow. For some, using AI means having a chatbot help write emails and summarize meetings, while for others, it’s building custom agents, automating workflows, or generating in minutes what might otherwise take hours to build manually in Excel or other programs. Adding to the disparate meanings is the fact that in the pharmaceutical and biotech world, “AI” has been used interchangeably with terms like machine learning models, translational insights, digital twins, or advanced simulation environments.

None of these perspectives are wrong, they simply reflect where each person sits within a rapidly evolving AI ecosystem. However, with growing pressure on individuals and teams alike to expand AI skills, lacking a shared definition creates a risk of building disconnected capabilities rather than connected intelligence.

I have seen this over the last decade or more with companies looking to upskill in their modeling and simulation platforms to support translation discovery and clinical research. It requires a coordinated effort to allow teams to work together to implement the best solution for their company. Similarly, everyone is now racing to upskill in AI and develop their own interpretation of what that means without always considering how those technologies communicate, integrate, or support the wider scientific process within their companies.

That seems like a missed opportunity.

Right now, organizations have the chance to incorporate AI in an ecosystem framework, where technologies, data, scientific models, and people work together seamlessly. Efficiency in our industry comes from creating environments where information flows freely, where insights can be trusted across teams and disciplines, and where AI enhances decision-making rather than complicating it.

However, to make this a reality, leaders must first create and communicate a shared vision of the purpose and goals of AI use within their organizations. Otherwise, they will continue to be plagued by inconsistent and duplicative applications and agents, running up unnecessary costs without realizing the full benefit their technologies can offer.

Here is my vision for your consideration, a vision that is shared by all of us at Simulations Plus: I believe the future of AI in modeling and simulation will be defined by connectivity and interoperability. We need platforms and ecosystems that bring together mechanistic science, simulation technologies, data-driven AI, and human expertise into a coherent framework.

AI should not sit outside the scientific process as a separate layer of technology. It should exist within it, acting as a co-scientist, supporting and enabling better decisions across discovery, development, and clinical research.

The organizations that succeed over the next few years will not simply be the ones with the largest datasets who are adopting AI the fastest. They will be the ones that build connected ecosystems capable of turning complex science into faster, more confident decision-making and ultimately delivering better therapies to patients in a reliable and trusted way.

If you are interested in learning more about how to implement an AI-powered ecosystem, I’m happy to chat.

Schedule a Call

The evaluation of a drug candidate’s pharmacokinetic (PK) properties, while critical for the success of a drug, has traditionally been excluded from the early drug development process, in large part due to the cost of assays and the difficulty predicting PK properties from structures. Instead, early drug discovery efforts rely on in vitro assay data as surrogates for the in vivo PK properties. Including downstream PK information at the point of design could expedite drug development, delivering leads with desired PK properties from the earliest stages of development. Several proprietary, commercial, and open-ware platforms have been developed to rapidly predict PK properties, and they are beginning to be incorporated into early drug design. However, evaluating the performance of these “high-throughput” PK prediction platforms has been challenging due to the lack of publicly available, well-curated PK datasets for benchmarking. We assembled a machine-readable database of plasma concentration–time profiles spanning 180 compound–species–route–dose cases, five species, and three routes of administration. The database contains standardized chemical structures, dose and study metadata, digitized concentration-time observations, uncertainty bounds where available, and consistently derived PK endpoints. We demonstrate its use by evaluating high-throughput mechanistic PK workflows parameterized with predicted or measured inputs. Whole-curve error metrics complemented conventional endpoint comparisons and enabled analysis of model limitations, uncertain inputs, and curation errors. The results support concentration–time profiles as an informative basis for reproducible PK benchmarking and show how a shared benchmark can guide model development rather than serve only as a leaderboard. This openly maintained resource is intended to support transparent, reproducible comparison and continued improvement of high-throughput PK prediction methods.

By Jeremy O. Jones, Rafał A. Bachorz, Michael S. Lawless, David W. Miller, Robert Fraczkiewicz, Viera Lukacova

Drug-induced liver injury (DILI) is a major cause of acute liver dysfunction and remains one of the most challenging adverse drug reactions encountered in clinical practice. Epidemiologic studies estimate an annual incidence of 14–19 cases per 100,000 individuals, with DILI accounting for approximately 10% of acute hepatitis presentations and a substantial proportion of acute liver failure in Western countries. (1, 2) Clinical manifestations range from asymptomatic elevations in aminotransferases to fulminant hepatic failure, and the unpredictable nature of idiosyncratic reactions continues to complicate diagnosis and risk assessment. (1) The severity and continually evolving nature of DILI has pushed the field toward developing predictive approaches that can improve patient safety, avoid late-stage development failures, and prevent costly postmarket withdrawals. This theme was highlighted at the “Predicting Drug-Induced Liver Injury” symposium at the ACS Fall 2025 conference, where the discussions centered on how traditional in vivo and in vitro toxicology methods can be integrated with modern in silico tools. Together, these complementary strategies underscore the growing emphasis on building more reliable and mechanistically informed DILI prediction frameworks.

By Don Pivithuru Liyanarachchi

The expiration of the HSR waiting period satisfies one of the regulatory conditions necessary for the completion of the transaction. The transaction remains subject to the satisfaction or waiver of other customary closing conditions, set forth in the merger agreement, including approval by Simulations Plus shareholders and receipt of certain regulatory approvals in France. Subject to the satisfaction or waiver of the remaining closing conditions, the transaction is currently expected to close in the second half of calendar 2026.

 
By Keyla Nunes Farias Gomes, Raíssa Maria dos Santos Galvão, Natalia Lidmar von Ranke, Carlos Rangel Rodrigues, Caroline de Souza Ferreira Pereira, Julianne Soares Pereira, Matheus Amorim Rosa e Silva, Brenda Bairral Queiroz Ornellas, Jonathas Albertino de Souza Oliveira Carneiro, Geovana Espindola Jardim, André Lopes Fuly, José Augusto Albuquerque dos Santos and Robson Xavier Faria

That trust is not driven by any single output or diagnostic, however; it emerges from the consistency between the modeling objective, the data, the assumptions, and the conclusions.

There are key elements that help form that consistency and build regulatory trust. This blog post details five that are critical to consider as you plan and implement your development program.

1. Clear alignment with the regulatory question
   

   The model is explicitly designed to address the decision of interest (e.g., dose selection, labeling), with a logical connection between objectives, data, and conclusions.

2. Validation aligned with model risk and impact
   

   The extent of model evaluation reflects its role in decision-making; high-impact models require more rigorous testing, including sensitivity analyses and alternative scenarios.

3. Reproducible workflows
   

   Model code, datasets, and execution steps are sufficiently documented to allow independent re-analysis without ambiguity or manual intervention.

4. Transparent assumptions and limitations
   

   Structural choices, covariate relationships, and data handling decisions are clearly justified, with an explicit discussion of their potential impact on conclusions.

5. Consistency between diagnostics and conclusions
   

   Goodness-of-fit, predictive checks, and other diagnostics support the stated conclusions, with no unresolved discrepancies. Ultimately, regulatory confidence is built when a model is not only technically sound, but understandable, reproducible, and clearly connected to the decision it informs.

Transparency, interpretability, and validation are not independent concepts. They work together to establish trust. When aligned, they enable models to serve their intended role: supporting decisions that impact patient safety and therapeutic outcomes.

If you’d like to enhance trust in your program, schedule a 15-minute call with our regulatory experts to learn where you can optimize your efforts.

Schedule a Call