Modeling & simulation and artificial intelligence for early drug development

In early drug development, the value of modeling & simulation (M&S) and artificial intelligence (AI) lies not in their application per se: the critical skill is knowing when, how, and at which decision points to deploy them to inform smarter development strategies.

The pharmaceutical industry has reached a tipping point driven by rising protocol complexity, tighter capital, fluid regulatory expectations, and unrelenting pressure to tell a cohesive, data-driven value story early. In this context, modeling and simulation (M&S), together with emerging artificial intelligence (AI) tools, are crucial in reshaping how development programs are designed, executed, and interpreted. However, it is important to distinguish their roles, separate hype from reality, and understand how these approaches intersect to maximize the benefits from these tools.

M&S has been integral to drug development for decades and is well established in regulatory decision-making. Recent regulatory developments, represented by the ICH M15 Guideline on general principles for model-informed drug development (ICH M15), effective July 23, 2026, harmonize expectations and recommendations for model-informed drug development (MIDD), in terms of planning, model evaluation, documentation of evidence, and related regulatory interactions. AI is a more recent addition to drug development that primarily accelerates processes, acting as an assistant and supporting the partial or full automation of workflows.

M&S & AI are powerful approaches in their own right and can be even more impactful when used together. Maximizing their value, however, requires understanding which approach, or combination of approaches, is best suited to the specific scientific question and development decision. Applied thoughtfully at the right points across early drug development, these tools can generate stronger evidence, enable better-informed decisions and reduce overall development time. 

1.    M&S to drive better decisions and stress-test failure early

The most effective development programs integrate M&S from the earliest stages, ensuring that each study contributes to a coherent and evolving evidence base. M&S generates value by integrating data and knowledge to inform decisions, such as selecting and justifying doses, supporting study design, and defining the patient population most likely to benefit from treatment. The role of M&S extends beyond assessing whether a drug may work to characterizing uncertainty, evaluating alternative hypotheses, and identifying the factors most likely to determine clinical success. The greatest value is realized when the right M&S approach is applied to the right question at the right decision point in development.

Importantly, M&S is not a one-time exercise, but a continuous component of drug development. Different modeling approaches are applied as new data sources become available and new questions arise: from early translational models leveraging preclinical findings to support first-in-human (FIH) dose selection to clinical models integrating human pharmacokinetics, target engagement, efficacy, and safety data. While models may be refined or new M&S approaches might come into play over time, the underlying role of M&S remains constant: providing quantitative insight to inform development decisions.

Example: The versatility of M&S is illustrated by a sponsor program in which we assessed an alternative dosing strategy: transitioning from body weight-based dosing to a fixed-dose regimen. By leveraging existing clinical data and population PK/PD models, simulations demonstrated comparable exposure, safety, and efficacy, supporting the alternative dosing strategy without requiring additional trials. This reduced complexity and improved usability while maintaining confidence in the benefit–risk profile. 

Timing is critical: early and well-integrated M&S approaches can transform development. Engaging modeling expertise early (ideally some time before FiH) allows for a comprehensive assessment of available data, identification of gaps, and development of a robust translational strategy. 

Example: In various FIH projects, we have used preclinical data to build a translational PK model and identify the exposure thresholds associated with the required pharmacological effect and with potential toxicity, respectively. Model simulations were then used to predict clinical doses, enabling the design of a FIH study that efficiently explored safety, PK, and (when feasible) target engagement in a wide range of doses.

For assessments such as drug-drug interactions (DDI), hepatic or renal impairment impact on exposure, or QT prolongation, M&S is a powerful tool to predict risk and guide study design. In this context, M&S is complementary to, and in some cases a substitute for, clinical evaluation. 

Example: In a Phase 2 program in which patients were expected to receive concomitant medications affecting enzymes important for the metabolism of the investigational medicinal product (IMP), Parexel applied M&S to predict the drug-drug interactions and their impact on IMP exposure. By integrating preclinical and early clinical data, the analysis demonstrated that certain concomitant therapies would not meaningfully alter exposure, enabling broader inclusion criteria in the Phase 2 trial. Therefore, this approach avoided unnecessary restrictions, easing recruitment while maintaining patient safety.

The regulatory weight of a model depends on both its technical sophistication and its context of use. While robust models can provide valuable insights in well-understood settings (such as a well-known drug class, mechanism of action, and/or intended target), strengthened by prior knowledge and experience, M&S is also essential in novel settings, where it can help structure available evidence, evaluate assumptions, characterize uncertainty and guide development decisions as knowledge evolves.

Although M&S can be applied throughout development, earlier integration generally provides greater opportunities to influence strategy and reduce uncertainty. Establishing a quantitative framework from the outset enables evidence-based decisions on dose selection, study design, and development planning, creating a stronger foundation for subsequent clinical and regulatory milestones.

2.    AI to accelerate study design, execution, and data contextualization

AI can support development decisions by rapidly identifying patterns, synthesizing, and extracting insights from large volumes of scientific, clinical, and regulatory information. At Parexel, we leverage proprietary AI, supervised and enriched by human medical expertise, to evaluate study designs and assess proposed endpoints against regulatory expectations. Expert oversight remains essential to interpret findings, ensure scientific validity, and place AI-generated insights within the broader context of the development program.

More in detail, we have deployed AI in early development to optimize study design, support site and patient selection, and expedite study start-up processes, including statistical programming, site contracting, grant management, and payments. 

For example, AI-driven tools can analyze large sets of historical protocols and published literature, identify patterns in endpoints, eligibility criteria, and trial designs, highlight potential inconsistencies with regulatory expectations, and suggest design optimizations based on prior evidence. These analyses rely on publicly available and appropriately governed data sources, ensuring that proprietary client information remains protected.

The true value of AI lies not only in the speed and scale of its analyses, but in its ability to augment scientific expertise. When applied thoughtfully and interpreted within the appropriate clinical and regulatory context, it can generate meaningful insights and support more informed development strategies.

Maximizing value through AI and M&S approaches

AI and M&S each provide unique capabilities that can strengthen drug development when applied thoughtfully and in the appropriate context. Their greatest impact is achieved when they are integrated throughout the development journey, from early translational planning through clinical development, to address evolving scientific, clinical, operational, and regulatory questions.

The most successful development programs do not treat AI and M&S as isolated activities. Instead, they deploy the right approach at the right time, leveraging quantitative evidence to inform study design, dose selection, patient population strategies, evidence generation plans, and other key development decisions as new data emerge. Insights generated by one approach may also inform and strengthen the application of the other, creating a more robust foundation for decision-making. Ultimately, the value of both AI and M&S derives from their thoughtful integration into a development strategy, the quality of the underlying data and evidence, and the expertise used to interpret and act upon the insights they generate. When embedded early and applied continuously, these approaches can help reduce uncertainty, support better-informed decisions, and accelerate the delivery of new therapies to patients.

Start a conversation

Continue exploring Phase I Strategies

Gain practical perspectives from the recent panel discussion featuring biotech leaders and early phase experts on the strategies that can accelerate proof of concept.

Watch on-demand

Explore more episodes