CytoReason

The problem: drug development is expensive and prone to failure

Developing a new drug is a long and expensive process. In the industry it is seen as a process that can take more than a decade and cost hundreds of millions to billions of dollars for a single drug, and many clinical trials fail because the drug does not work as expected in the human body, even though it succeeded in early lab tests. CytoReason, founded in 2016, tackles this gap by building computational models that simulate how a disease, such as cancer, inflammatory bowel disease or an autoimmune disease, works at the level of the single cell. The models are based on huge amounts of existing research data collected in academic and clinical studies over many years.

The computational model as a partial substitute for an early trial

The models let drug companies check at an early stage, before an expensive trial in humans, whether a certain drug is likely to affect the disease and in what way. This saves time and money on development directions with low chances, and sometimes also helps identify new drug targets that were not clear in advance.

The expanded collaboration with Pfizer

CytoReason has reported long-term collaborations with several of the world's largest drug companies. The collaboration with Pfizer began in 2019, and on 20 September 2022 the companies announced its expansion: Pfizer would make a $20 million equity investment, with options to license CytoReason's platform and disease models and additional research funding, in a deal with a potential value of up to $110 million over five years. "Pfizer has been a strategic partner of CytoReason since 2019, and we are thrilled to scale our collaboration," said David Harel, CEO and co-founder of CytoReason, in the joint announcement.

Digital biology as a wider trend

CytoReason's approach sits at the meeting point of computational biology and artificial intelligence, a field sometimes called "digital biology". It represents a wider direction in which drug companies try to reduce their dependence on expensive physical experiments in the earliest stages of development, and to replace a large part of them with computer simulation based on real data.

An advantage over a general language model

CytoReason's medical-computational model has a built-in advantage over a general AI model. Instead of learning from free text on the internet, it is built specifically around structured molecular biology data, such as gene expression levels and interactions between immune system cells and different body tissues. So it can give more specific predictions for a given disease than a general language model that was not specially trained for this.