How AI Can Accelerate Clinical Trials' Probabilities Of Technical And Regulatory Success (PTRS)

By Kelly H. Zou, Ph.D., PStat®, FASA, CEO, AI4Purpose Inc. and president, American Statistical Association NYC Metropolitan Area Chapter

In biopharmaceutical R&D, probability of technical and regulatory success (PTRS) is a critical metric used to evaluate the likelihood that a drug candidate will successfully complete a randomized controlled trial (RCT) and obtain regulatory approval. It measures whether a drug candidate will both succeed technically to meet its clinical trial endpoints (efficacy, safety, etc.), as well as gain regulatory approval from agencies such as the FDA or EMA.

PTRS combines two distinct probabilities: probability of technical success (PTS), which assesses the chance of meeting clinical endpoints in the trial, and probability of regulatory success (PRS), which assesses the probability that a drug candidate will both succeed in its clinical trials and receive regulatory approval. It’s typically summed up as PTRS=PTS×PRS, where an example of PTS is meeting endpoints in trials, while an example of PRS is receiving an approval from a regulatory agency.

Pharma companies use PTRS to guide investment decisions, prioritize pipeline assets, and evaluate risk across development stages. PTRS can be dynamic and evolve as new data emerge. Specifically, interim analyses, competitors’ outcomes, and regulatory guidance updates may all influence the PTRS score. For example, a promising Phase 2 readout may increase PTS, while a change in FDA policy could impact PRS. For PRS, the strength of data, regulatory precedent, quality of submission, and engagement in discussion with the regulatory authorities may all play

Such adaptability makes PTRS a useful forecasting tool, especially for oncology or rare diseases.

Thus, PTRS can function as a strategic decision-making tool throughout the R&D lifecycle by:

• Supporting objective and data-driven decisions.
• Informing portfolio prioritization, budget allocation, and risk mitigation.
• Helping determine whether an RCT is worth pursuing, e.g., via design robustness, endpoint clarity, success rates in similar indications, and regulatory precedent.
• Encouraging transparency and consistency across teams, enabling cross-functional alignment between multi-stakeholders, such as clinical, regulatory, and commercial.

AI’s Impact on PTRS

AI is transforming how biopharmaceutical companies assess PTRS to evaluate whether a drug candidate can successfully pass the trial phases and gain regulatory approvals in the future.   

• Risk Assessment and Prediction: Machine learning (ML) and deep learning (DL) models help analyze vast datasets from past clinical trials, regulatory decisions, and biological data to estimate PTRS more accurately. AI helps identify patterns and correlations that human analysts might miss to help decision-making.
• Portfolio Optimization: AI tools allow companies to continuously update PTRS scores, as new data emerges, helping them pivot quickly if a drug’s prospects change. This dynamic insight supports better resource allocation, asset prioritization, and long-range planning across the drug portfolio.
• Accelerated Drug Development: AI speeds up target identification, compound screening, and efficacy prediction, which feeds into early PTRS estimates. By predicting which compounds or a class of compounds are the most likely to succeed, AI helps avoid costly failures and focus on high-potential candidates, while expediting the trial process.
• Drug Repurposing and Expansion: AI helps uncover new indications and expand product labels for existing drugs, increasing their PTRS for alternative treatments.
• Bias Reduction: AI introduces rigor into what were traditionally subjective assessments, helping reduce bias in go/no-go decisions through scenario planning and simulating various trial outlooks.

AI In R&D

There have been evolving trends of AI in R&D, including:

• Accelerating Drug Discovery: One of the most profound impacts of AI in R&D is its ability to dramatically speed up the discovery process. Traditional R&D often involves labor-intensive experimentation and trial-and-error approaches. AI flips that model by using predictive algorithms and machine learning to simulate outcomes, identify promising candidates, and optimize experimental design. For example, AI efficiently analyze massive datasets of molecular structures and biological interactions to predict which compounds are most likely to succeed in clinical trials. This not only shortens the timeline from lab to market but also improves PTRS. In addition, decentralized clinical trials (DCTs), pragmatic clinical trials (PCTs), and real-world evidence (RWE) studies may provide additional quantitative insights that can be complimentary.
• Making Data-Driven Decisions: AI enhances decision-making by providing data-driven insights that go beyond human intuition. Natural language processing (NLP) tools scan millions of research papers, patents, and clinical trial reports to extract relevant information and identify emerging trends. This helps R&D teams stay ahead of the curve and avoid duplicating efforts. Moreover, AI-powered platforms model complex systems, e.g., climate simulations, chemical reactions, or supply chain dynamics, allowing researchers to test hypotheses virtually before committing resources to experiments.
• Transforming Experiments: AI is also transforming how experiments are designed and executed. Robotic labs powered by AI autonomously conduct thousands of experiments, analyze results in real time, and adjust parameters accordingly. Experiments are particularly valuable in fields like pharmaceutical manufacturing, including fixed dose combinations.
• Utilizing Diverse Data Sources: R&D generates big data, much of which goes underutilized, due to 4Vs, i.e., volume, velocity, variety, and veracity. AI excels at mining this data for insights. ML algorithms detect subtle patterns and correlations that humans might overlook, leading to unexpected discoveries and new lines of inquiry. For example, in genomics, AI can analyze gene expression data to uncover links between genetic markers and disease susceptibility. In aerospace, AI models predict material fatigue and failure rates, improving safety and performance.
• Fostering Collaborations and Partnerships: AI is breaking down silos between disciplines, enabling more integrated and collaborative R&D efforts. Platforms that combine AI with cloud computing and digital twins allow researchers from different fields collaborate and share data rapidly, heading towards real-time.
• Following Ethical Standards: Data quality and bias can skew results, and overreliance on algorithms may lead to missed opportunities or ethical gaps. Transparency, explainability, and human oversight remain critical to ensure responsible innovation.

Summary

AI is reshaping R&D in the biopharma industry. It is accelerating discovery, reducing costs, and enabling breakthroughs that were previously unimaginable. There are innovative ways how RCTs are designed and conducted aided by AI. It is important to carefully examine the strength and quality of trial data, regulatory precedents and evolvement, quality of submission packets, automating literature and documentation processes via LLMs, and early and systematic engagement with regulatory bodies.

Beyond PTRS, PTS, and PRS, there are also other probabilistic metrics, such as probability of success (POS), likelihood of approval (LOA), probability of phase transition (POPT), etc. Each probability, which compliments the PTRS, may help lead to better decisions to optimize R&D, where AI plays both timely and impactful roles.

Note: The content of this article may not reflect the opinions of the author’s affiliations.

About The Author:

Kelly H. Zou, PhD, PStat, FASA, is CEO, AI4Purpose Inc. She is the president of American Statistical Association’s New York City Metropolitan Area Chapter. Previously at Pfizer Inc, she was vice president, head, medical analytics and insights; senior director, real world evidence; group lead of methods and algorithms; analytic science lead; senior director, statistics lead. She was head, global medical analytics, real-world evidence, and health economics and outcomes research, Viatris Inc. (merged from Pfizer). Earlier, she was Associate professor, radiology, and director, biostatistics, Harvard Medical School and affiliated hospitals. She was associate director, rates, Barclays Capital. She received both an MA and a Ph.D. in statistics from the University of Rochester. She was the winner of multiple global awards, including Chief Data and Analytics Officers’ Forum’s Future Thinking, AI100, Top 100 Data & Analytics Professional, Reuters Events Pharma USA’s Most Valuable Data & Insights Initiative Team, as well as a Reuters Tra