AI meets human tissue to fast-track precision medicine development

By combining human tissue models with explainable AI, researchers can analyse complex patient data to identify which treatments work best for which patients. First applied to inflammatory bowel disease, this approach could improve clinical trial success rates across many diseases.

A collaboration between REPROCELL, IBM and the STFC Hartree Centre has created an AI-powered platform that combines human tissue models with machine learning (ML) to streamline clinical trials, cut costs and accelerate the delivery of new medicines.

Delivered through the Hartree National Centre for Digital Innovation (HNCDI), the project is helping shape the future of precision medicine by tackling one of drug development’s biggest challenges – predicting individual patient responses to treatments. We spoke with Graeme Macluskie of REPROCELL, Dave Braines of IBM, and Dan Clayton and Ruby George of the Hartree Centre to learn how this collaboration is bringing together artificial intelligence and human tissue-based models to make drug development faster, more targeted and more successful.

Decoding complex patient data

Precision medicine relies on understanding why individuals react differently to the same therapy. These differences can come from sources such as genetics, lifestyle and environmental factors – creating datasets too large for traditional analysis. Graeme Macluskie, Director of Precision Medicine at REPROCELL, explained why AI was the logical solution.

“Understanding the underlying reasons why individual patients can respond differently to drug treatments is really the ultimate goal that underpins effective precision medicine,” he said. “Finding trends within these vast datasets is obviously a huge challenge. It was something that we thought (and have since shown) that AI/ML could help with.”

The AI platform, known as Pharmacology-AI, does not replace scientists but enhances their ability to detect patterns. Dave Braines, IBM’s technical lead on the project, highlighted its role as a decision-support tool.

“For me, the key value of platforms like Pharmacology-AI lies not in generating definitive answers or making decisions, but in accelerating and augmenting human expertise,” he noted. “Their true potential is in significantly boosting the efficiency of experts – within this domain and beyond.”

Dan Clayton, Research Software Engineer at the Hartree Centre, added that interpretability was just as crucial as predictive power, said:

“One of the most important aspects about this project was to make the machine learning outputs interpretable. If we had a model that could accurately predict a person’s drug response but not give us any insight as to how it got there, it would not be of any use for clinical trial design.”

Ruby George, also a Research Software Engineer at the Hartree Centre, worked on the project’s dashboard – the key interface for end users, shared:

“From a software engineering perspective, the main challenge was minimising latency when loading the various graphs on the dashboard. Because these visualisations came from high-volume data, we had to optimise both backend data access and frontend rendering to ensure a smooth, responsive user experience.”

Leveraging human tissue models

At the centre is REPROCELL’s human fresh tissue models – which preserve the biological complexity of real patient samples. These models allow researchers to simulate drug effects before clinical trials, generating data that is both reliable and highly relevant.

“Human fresh tissue models utilise tissue residual from surgical procedures or non-transplantable organs and allow the effects of drug treatments to be measured in a laboratory environment. Data from these assays is extremely valuable and is really the only way to generate the depth of data necessary for most precision medicine applications, prior to the clinic.”

– Graeme Macluskie, Director of Precision Medicine, REPROCELL

To integrate these models with AI, the project adopted an agile development approach and close collaboration across all partners.

“The project tackled a complex and multi-dimensional problem space and the final solution reflected a high level of technical rigour and thoughtful design. Each conversation and demonstration laid the groundwork for further enhancements and a truly robust outcome.”

-Dave Braines, CTO for Emerging Technology, IBM

Clayton emphasised the importance of usability, ensuring that even non-technical staff could interpret and act on the data, saying:

“We worked on integrating clinical metadata to ensure that the visualisations in the platform are easy to use, even for non-technical staff and ensure that the ML outputs are explainable.”

Expanding the impact

The platform was initially applied to inflammatory bowel disease (IBD) but has potential to be used for other conditions. By using high-quality patient data, it could be adapted to a wide range of therapeutic areas.

“The overarching concept of this project and the Pharmacology-AI technology is broadly applicable to any disease area. Currently, our primary focus is IBD, but we are assessing the market demand and clinical need for other clinical and preclinical applications.”

– Graeme Macluskie

Dave Braines believes this technology will become an indispensable part of research and development across the life sciences, sharing:

“By surfacing patterns, insights, or detailed correlations, platforms like this can act as powerful allies to human experts – delivering greater speed, depth and clarity.”

Transforming drug development

By identifying optimal patient profiles early, the platform could reduce costly trial failures and speed up drug approval.

“Improving the attrition rate of Phase II and Phase III clinical trials by 10 percent has the potential to reduce the average capitalised cost of getting a drug to market by hundreds of millions of dollars.”

– Graeme Macluskie

The collaboration between REPROCELL, IBM and the STFC Hartree Centre demonstrates how the combination of advanced technologies and domain expertise can develop the future of healthcare. By combining human tissue models with the analytical power of AI, the team has created a platform that streamlines drug discovery and improves patient outcomes.

As this technology evolves, it promises to expand into new disease areas, find hidden insights within complex datasets and ultimately bring more effective treatments to patients faster.

“With thoughtful oversight, we see vast potential for applying this approach across a broad range of R&D scenarios – and this is just one of them.”

-Dave Braines