Pfizer, CytoReason extend AI pact after immune system success

Pfizer has extended its collaboration with CytoReason, positioning it to continue using the computational disease models in drug development.

CytoReason disclosed its original partnership with Pfizer in 2019. Back then, Pfizer committed to pay up to “low double-digit millions of US dollars​” for access to technology designed to address a problem that has emerged as the pace of data generation has accelerated.

“One of the major problems we’re addressing is what we call the data-insight gap in pharma R&D. Namely, that the volume of human molecular data is growing exponentially, while our analytical capabilities are growing linearly. To us, it’s clear that humans alone cannot process ever-growing amounts of data and close the gap​,” said David Harel, CEO and co-founder of CytoReason.

Under the 2019 deal, Pfizer has used CytoReason’s models to improve its understanding of the immune system, a key topic for everything from its record-breaking COVID-19 vaccine to its immuno-oncology aspirations.

CytoReason created a platform with the potential to boost Pfizer’s understanding of the immune system by bringing together thousands of samples on a cell-protein-gene level and by seeking to address the data-insight gap through the use of machine learning.

“The platform we built, which is made up of computational disease models, is designed to digest, organize, and make sense of all data types and sources, from multiple indications and treatments. Together with our customers, we use the insights we extract to address a variety of use cases, such as prioritizing new targets, finding biomarkers, profiling combinations, and more​,” said Harel.

In the Pfizer collaboration, the computational models have demonstrated their potential to improve the chances of success of clinical trials. Specifically, the project has shown the models may be able to help identify which subpopulations of patients are likely to benefit from an investigational drug candidate.

“Computational models allow for a fast, efficient, and data-driven stratification to take place. Since our models are based on hundreds of patient samples, we can use them to detect sub-populations of patients prior to treatment initiation. Decision-making is then driven by clear sub-group characteristics and becomes personalized​,” said Harel.

Work is underway to expand the platform. As well as growing the platform by feeding in more public and proprietary data, CytoReason is adding more disease models that enable it “to compare molecular mechanisms across multiple diseases, patient groups and treatments​,” said Harel.