Advanced Computing in the Age of AI | Wednesday, December 6, 2023

UK Using Cognitive Computing in Personalized Medicine Push 

As the pharmaceutical industry moves from the era of “blockbuster” drug discovery (developing a single drug that targets large numbers of people with a particular disease) to the era of personalized medicine (tailoring drugs to the unique physiology of individual patients), so too has come growth in the use of big data analytics and digital drug design by the pharmaceutical industry.

In the UK, a four-year project leveraging big data analytics technologies brings together leading British universities, government agencies and major pharmas (Pfizer, GlaxoSmithKline, AstraZeneca and Bristol-Myers Squibb) to develop digital design techniques to streamline development and manufacturing of meds, enhance predictive processes and, ultimately, give patients faster access to the drugs that are right for them.

Adrian“This project is about taking existing APIs (active pharmaceutical ingredients) and optimizing the manufacturing process all the way from that molecule to the final drug product,” Adrian Toland, Business Development Manager at the U.K.’s Science & Technology Facilities Council (STFC), told EnterpriseTech. “In pharmaceutical manufacturing environments there are a number of unit operations, processes and procedures, and all of these will be generating vast quantities of data. These companies are data rich, the real challenge is extracting value from this data so that the products that are developed have got the specific performance attributes that are required by the manufacturer.”

At the heart of the project will be several HPC systems that reside at the STFC’s Hartree Centre “industrial gateway” providing HPC resources to companies, resources that include an IBM Blue Gene system comprised of 131,008 cores, 131TB memory; IBM Idataplex and Nextscale systems with 18,400 Intel Xeon cores, 48 GPUs and 41 Intel Phi processors. Toland said the centre in recent years has expanded from an emphasis on modeling and simulation into what IBM calls “data-centric computing.”

Toland emphasized that the project, called ADDoPT (Advanced Digital Design of Pharmaceutical Therapeutics), is not focused on developing APIs that target specific diseases but on developing tools and methodologies for specific patient stratifications, based on statistical analysis that groups patients for optimal drug therapies based on their genetic information or other molecular or cellular analysis.

‘Digital design’ combines research insight and mechanistic modelling to provide links between raw materials, formulation, manufacturing processes and drug product quality. It spans all operations, processes and procedures during the development and manufacture of medicines, and their application.

ADDoPT will bring together a range of predictive models and insight from industrial case studies at the four major pharmaceutical companies, allowing more targeted future experimentation, a better understanding of risk, and better design and scale-up for robust drug products and processes.

“The purpose of the project is to harness this vast quantity of data being generated at all stages in the drug manufacturing process,” said Toland, “and to use that data to develop predictive models on how to make drug products that have enhanced performance attributes. The reason we’re doing this now is because we have the technology to actually do this. We have essentially got huge amounts of data that is multi-variant. It’s impossible for a human being to identify significant patterns within these data sets, but we can do this with the data analytics technologies we now avail to us.”