Integrating AI and Machine Learning in Drug Discovery and Development

Pharmaceutical companies face a variety of challenges, such as the "patent cliff" and the difficulty of discovering new "blockbuster" drugs, when researching and developing potential new products (1). As part of the drug development toolbox, pharma companies are integrating AI and machine learning (ML) not only during discovery and development, but across the entire lifecycle of a drug (1,2). This technological shift is not merely about automation but reimagines how therapies are discovered, tested, and brought to market (1, 3).

A report from the Capgemini Research Institute details how the industry is moving to AI to reduce costs and clinical failure rates (4). A survey of 500 senior executives across eight countries found that a majority (82%) believe AI will fundamentally transform biopharma R&D, with 63% saying that a failure to scale AI will leave companies behind when it comes to innovation and market relevance (4). They also anticipate that in the next decade, most new molecular entities will be created using AI-driven platforms.

How is AI accelerating early-stage discovery and solubility?

Poor solubility of new chemical entities is consistently a challenge in drug development, with an estimated 70–90% of drug candidates categorized as poorly soluble (5, 6). Traditionally, identifying the right formulation for these "brick dust" or "greaseball" molecules required extensive and expensive empirical trial-and-error (5,6). However, in-silico modeling using AI and ML is now being leveraged to predict physicochemical properties and guide early-stage decision making (5,6). Researchers can identify optimal solubilization methods and suitable excipients by utilizing tools like quantum mechanical calculations and molecular dynamics simulations (5). This "ultra-material-sparing" approach significantly reduces API consumption and saves money in research costs (5). A notable example is the evaluation of the compound CVN424, for which predictive modeling guided the selection of spray-dried formulations, accelerating the overall development timeline (6).

How are digital twins improving personalized clinical evaluations?

The use of digital twins in preclinical evaluation is revolutionizing how drugs are tested on human organs (7). For example, by using an ex-vivo lung perfusion system, researchers can keep human lungs alive outside the body to collect "clean" multimodal data. Machine learning models then use these data to create a personalized digital control arm for every organ being treated.

This approach allows for a direct comparison between observed treatment effects and a digital twin-generated, "untreated" outcome within the same organ. By reducing the reliance on large control groups and traditional animal models, digital twins may reduce study sizes and accelerate clinical trials (7).

How are regulatory workflows and market access streamlined?

New regulations in the European Union requiring complex joint clinical assessments has added to the burden of regulatory compliance and health technology assessments. Generative AI (GenAI) is being utilized to work with the raw data to complete these submissions (8) by automating the creation of clinical evidence summaries, reducing the time spent on initial dossier drafting by approximately 40% in some cases.

Beyond documentation, agentic AI can automate labor-intensive processes in trial master file management, intaking and classifying documents with "human-in-the-loop" checkpoints for quality control (9).

Domain-specific models and secure collaboration

While general-purpose AI models are proficient in language, they often lack the depth required for the highly regulated pharmaceutical market (9). To combat this, domain-specific language models can be trained on proprietary enterprise documentation, such as batch records and validated procedures to support high-stakes decisions regarding drug manufacturing and clinical protocols (10).

To improve communication between regulators and drug developers, secure cloud environments, like the PRISM project, are enabling both parties to work on the same documents in real-time, using AI agent-like functions to accelerate the regulatory process and ensure that safety and efficacy are addressed long before a product reaches the patient (1,2).

What AI tools are entering the drug development space?

In recent years, drug developers have put a lot of time, money, and effort into oncology. In January 2026, SOPHiA GENETICS, a company working in AI-driven precision medicine, announced a collaboration with MD Anderson Cancer Center, which will utilize SOPHiA GENETIC’s AI-powered analytics platform. The organizations will launch research and development programs and co-develope a next-generation sequencing oncology test to translate complex multimodal data (11).

Also in January, the AI-drug discovery company Oxford Drug Design announced that it has successfully completed in vivo validation of a novel therapeutic approach targeting multiple tumor types in the development of a potential first-in-class cancer therapy using its GenAI platform (12). “In studies using a genetically engineered mouse model that replicates the earliest mutational events in colorectal cancer, Oxford Drug Design’s lead compound demonstrated statistically significant anti-tumor activity with efficacy comparable to that of rapamycin–a benchmark therapy– while showing no detectable signs of toxicity,” the company explained in a press release (12).

Iktos, a company that applies AI and robotics to drug discovery, entered into a multi-target collaboration agreement with the pharmaceutical company Servier with the goal of leveraging Iktos’ AI-orchestrated discovery platform to accelerate the design and optimization of novel small-molecule therapeutics in oncology and neurology. Iktos will apply its generative AI and AI-orchestrated robotics platform to design, synthesize, and optimize small molecules for multiple undisclosed targets, which will then be evaluated and potentially selected by Servier for preclinical and clinical development (13).

The human element in a digital ecosystem

Experts stress that AI and other digital technologies are force multipliers, not replacements for human expertise (7,9). Successful integration requires a "digitally fluent" workforce who can supervise automated systems and interpret AI-generated impact assessments (3). By automating administrative burdens, AI allows scientists and regulatory experts to return to the core of their work: scientific reasoning and clinical judgment (2, 8).

Responsible deployment of AI and machine learning creates a more efficient and equitable path to innovation, shortening the time between the discovery of a molecule and the treatment of a patient (8).

References

1. Cole, C. Global Collaboration and Technological Acceleration to Evolve Drug Innovation. PharmTech.com, Dec. 24, 2025.
2. Cole, C. Harnessing Data and Secure Cloud Environments for Global Pharma Solutions. PharmTech.com, Dec. 23, 2025.
3. Cole, C. Orchestrating the Pharmaceutical Future: Living Decision Engines and AI Integration. PharmTech.com, Dec. 23, 2025.
4. Capgemini. Smart Bet, Only Option, or Both?: Biopharma R&D turns to AI. https://www.capgemini.com/insights/research-library/how-ai-will-transform-life-sciences-r-and-d/ (accessed Jan. 13, 2026).
5. Mirasol, F.; Lavery, P. Leveraging AI/ML to Reduce Risk in Drug Development: Part One. PharmTech.com, Nov. 18, 2025.
6. Cole, C. Using AI and Molecular Dynamics to Overcome the Poor Solubility Challenge. PharmTech.com, Nov. 10, 2025.
7. Cole, C.; Mirasol, F. Achieving a Personalized Digital Control Arm for Preclinical Drug Evaluation. PharmTech.com, Nov. 13, 2025.
8. Dhas, A. Accelerating HTA Readiness with Generative AI. PharmTech.com, Dec. 17, 2025.
9. Medable. Medable Launches New AI Agent for Automating Clinical Trial Document Management; Presenting at JPM Health. Press Release. Jan. 6, 2026.
10. Nair, J. Why Pharma’s AI Agents Need Smaller, Domain-Specific Models First. Pharmaceutical Technology, Dec. 3, 2025.
11. SOPHiA GENETICS. SOPHiA GENETICS and MD Anderson Announce Strategic Collaboration to Accelerate AI-Driven Precision Oncology. Press Release. Jan. 7, 2026.
12. Oxford Drug Design. Oxford Drug Design Announces Further In Vivo Validation of Novel Oncology Therapeutic Mechanism. Press Release. Jan. 12, 2026.
13. Iktos. Iktos Enters Strategic €1B+ Multi-Target AI Partnership with Servier to Advance Drug Discovery in Oncology and Neurology. Press Release. Jan. 8, 2026.

About the author

Susan Haigney is lead editor for Pharmaceutical Technology®.