Lindus Health, the anti-CRO running radically faster, more reliable trials, has been awarded funding by ARIA, the UK’s specialized R&D funding agency built to unlock scientific and technological breakthroughs, to design safe and reliable AI for clinical trial design as part of their Safeguarded AI program. The project called “Transforming clinical trial design using safe and reliable AI,” will be a collaboration between Lindus Health and Nobuko Yoshida’s research group in the Department of Computer Science, University of Oxford.
Safety and reliability of AI systems represents a critical barrier to innovation in safety-critical sectors like healthcare, where errors can cost lives and undermine public trust. Today’s AI systems, while powerful, lack the reliability needed for high-stakes applications such as clinical trial design, where hallucinations or biased decision-making could lead to undesirable outcomes. The Safeguarded AI programme aims to revolutionize AI safety by combining scientific world models with mathematical proofs to create AI systems with quantitative safety guarantees—like those seen in industries like aviation and nuclear power—to unlock transformative benefits in critical sectors while minimizing risks.
Lindus Health is pioneering a breakthrough AI-assisted clinical trial design system as part of this ARIA-funded project. The research will validate methods for using AI with mathematically proven safeguards to transform decision-making in trial planning. This collaborative work directly tackles the fragmented, inefficient processes that currently plague trial design, leading to errors and wasted historical data. The project aims to develop a solution that leverages mathematical proofing techniques to recommend best practices to clinical trial researchers while delivering verifiable safety guarantees. The innovation promises to substantially reduce trial setup times, lower costs, and increase accuracy—ultimately accelerating the development of life-saving therapies while building much-needed trust in AI tools among clinical researchers who have been reluctant to adopt them. Once fully validated, this approach could fundamentally reshape how the biopharmaceutical industry approaches clinical trials, making them faster, more reliable, and more accessible.
“Poor trial design can lead to various consequences that hinder scientific innovation, from time-consuming operational tasks to clinical trial failures” said Nijat Hasanli, Head of Product at Lindus Health. “With ARIA, we’re combining our AI and industry expertise to work towards eliminating these risks, setting every clinical trial up for success and transforming the way we develop medical interventions.”
“AI could unlock transformative improvements in our critical infrastructure, but right now adoption is limited because without ironclad safety assurances, we risk unintended and damaging consequences,” said David ‘davidad’ Dalrymple, Programme Director at ARIA. “Our goal is to prove that it’s possible to develop AI with quantitative safety guarantees, and that this could unlock significant economic value for the UK.”
Lindus Health has sustained a continuous role in driving and supporting AI innovation in clinical research. The company has been published in Nature for proprietary AI technologies to predict trial outcomes from historical data to inform trial design as well as generate trial protocols and documents to accelerate study startup processes. Lindus Health is actively partnering with The Clinical Data Interchange Standards Consortium (CDISC) to develop standardized templates for sharing and submitting clinical trial data, an initiative that has the potential to elevate the FDA’s standards for research data. This work along with its other AI use cases has been recognized in the Pinnacle Healthcare Awards and the Sifted 100’s fastest-growing startups in the UK, and presented at the CDISC + TMF Europe Interchange and CNS Summit.
Explore AITechPark for the latest advancements in AI, IOT, Cybersecurity, AITech News, and insightful updates from industry experts!