by Clinmark

In our previous article, we explored the AI Act and its impact on the clinical trials industry in Europe. This time, we would like to move from the regulatory outlook to practice and take a closer look at how artificial intelligence is already reshaping research today. 

From the discovery of novel molecules such as rentosertib, designed entirely with generative AI, to the use of machine learning models in trial design, recruitment, and endpoint assessment, the field is moving from theory to real-world application. Importantly, regulators are starting to recognize and validate AI-based tools, setting important precedents for their wider use. 

This article reviews the most recent breakthroughs, real-life examples, and practical applications of AI in clinical trials.  

What’s New in Drug Discovery? 

Only a few years ago, artificial intelligence was primarily viewed as a supportive tool for molecule design. Today, we already see drug candidates that would not exist without it. 

• Rentosertib (ISM001-055, Insilico Medicine) 

One of the most notable examples is rentosertib, a TNIK inhibitor developed entirely with the use of generative AI models. Algorithms not only helped identify a novel therapeutic target but also designed the candidate molecule itself. Rentosertib is currently being evaluated in clinical trials for idiopathic pulmonary fibrosis (IPF), and the first Phase IIa results published in Nature Medicine indicate a favorable safety profile and promising efficacy signals. This represents the first case where both the molecular target and the compound structure were created with AI. 

• DSP-1181 (Exscientia + Sumitomo Pharma) 

Another milestone is DSP-1181, developed for obsessive–compulsive disorder. This compound advanced to Phase I clinical trials in record time, less than 12 months from the start of its design process.  

• The Expanding “AI-First” Pipeline 

Beyond these high-profile examples, a growing number of biotechnology companies are building their pipelines with an AI-first approach. This trend illustrates that AI is no longer just an add-on to traditional R&D, but increasingly the foundation of modern drug discovery. 

AI Across the Clinical Trial Lifecycle  

While AI-driven discovery makes headlines, its true impact may be even greater across the broader clinical trial process. From protocol design to recruitment, monitoring, and endpoint analysis, AI is becoming an integral part of daily operations, offering efficiency and new levels of precision. 

• Trial Design and Recruitment 

Machine learning algorithms are used to refine inclusion and exclusion criteria, identify patient subgroups most likely to benefit, and even support adaptive trial designs. This reduces amendments, minimizes screen failures, and accelerates timelines. 

• Monitoring & Safety 

AI-driven risk-based monitoring detects anomalies, predicts data quality issues, and improves resource allocation. In pharmacovigilance, predictive analytics help identify safety signals earlier and support more robust risk–benefit assessments. 

• Endpoint Assessment and Digital Biomarkers 

Regulators have validated AI-powered platforms for endpoint assessment, such as histopathology in MASH trials. Digital biomarkers, including wearable-derived data, are rapidly advancing through machine learning. 

• Synthetic Control Arms 

By leveraging historical and real-world datasets, AI enables synthetic control groups, reducing reliance on placebo arms and improving trial feasibility, particularly in rare diseases. 

Regulation and Validation 

For AI to gain trust in trials, rigorous validation is essential. Regulators expect clear documentation of training data, model architecture, and performance. They also stress reproducibility, explainability, and strategies to mitigate bias. Black-box systems without proper validation are unlikely to be accepted. 

Regulatory Precedents 

The European Medicines Agency (EMA) has recognized AI-based histopathology tools for endpoint assessment in NASH/MASH trials, demonstrating that regulators are open to AI integration when supported by robust validation. Similarly, the U.S. Food and Drug Administration (FDA) has issued guidance documents on the use of real-world data and digital health technologies, many of which are powered by machine learning. 

Audits and Oversight 

Sponsors and CROs must be prepared for audits that assess not only clinical data quality but also the governance of AI tools used in trial conduct. This includes evidence of algorithm validation, monitoring procedures for model drift, and processes for corrective action. As AI becomes central to trial operations, these audits will play an increasingly important role in regulatory interactions. 

Global Implications 

With the EU leading the way through the AI Act, regulators worldwide are shaping their own frameworks. For global trials, sponsors will need harmonized strategies that balance regional requirements with operational efficiency.  

Conclusion 

Artificial intelligence is no longer a distant promise. It is already reshaping how new therapies are discovered and how clinical trials are designed, conducted, and evaluated. From AI-discovered molecules like rentosertib, to predictive recruitment models, validated digital biomarkers, and synthetic control arms, the industry is moving rapidly from proof-of-concept to tangible impact. 

At the same time, progress comes with responsibility. Transparency, validation, and strong oversight must remain central to AI adoption. Regulators, sponsors, and CROs will need to collaborate closely to ensure these tools deliver not only speed and efficiency but also scientific integrity, patient safety, and lasting trust.  

References:  

Xu Z. et al., A generative AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis, Nature Medicine (2025). 

Insilico Medicine press release / pipeline pages re: Rentosertib (ISM001-055). https://insilico.com/tpost/tnrecuxsc1-insilico-announces-nature-medicine-publi 

Sumitomo Dainippon Pharma / Exscientia – DSP-1181 announcement. https://www.sumitomo-pharma.com/news/20200130.html 

Nature Biotechnology editorial / review on AI in clinical trials (2025) https://www.nature.com/articles/s41587-025-02754-1 

European Medicines Agency. (2025). EMA qualifies first artificial intelligence tool to diagnose inflammatory liver disease (MASH) in biopsy samples. https://www.ema.europa.eu/en/news/ema-qualifies-first-artificial-intelligence-tool-diagnose-inflammatory-liver-disease-mash-biopsy-samples