Recursion Pharmaceuticals: AI-enhanced drug discovery
Machine intelligence will transform every industry it touches. Healthcare is no exception, but complicated market dynamics make it difficult for new technologies to find adoption within giant provider systems where incentives are so woefully misaligned. Add to this the heaps of evidence required to gain regulatory approval and reimbursement and it’s no wonder we have yet to see AI bring substantive change to the practice of medicine. Computer vision outperforms human vision on most classification tasks, yet doctors still make diagnoses. For practical and ethical considerations, it will take some time for true AI-in-the-loop medicine to arrive.
Where we see AI having its first huge impact on healthcare is in drug discovery and development. Basic biology is so poorly understood and the interaction space is so vast, it is surprising that any drugs actually work. Combining traditional hypothesis-driven science with data-first methods will yield important discoveries that either alone cannot achieve. We see a huge opportunity for startups that can create massive troves of unique, well-labeled data and use them to build proprietary models of health and disease.
There is no better example of this idea in action than Recursion Pharmaceuticals. Today, we are proud to announce our investment in Chris, Blake, and the amazing company they have built. They have developed a highly generalizable strategy for applying computer vision to drug discovery that incorporates high throughput screening, automation, and machine learning. Their platform improves the more data they generate.
Recursion’s approach is fascinating. They use gene editing to create unique cellular models of genetic diseases. These cells are visually distinct from healthy cells. They then screen hundreds and thousands of small molecule drugs against the disease cells to see whether any of the drugs recover the healthy phenotype, i.e. make the disease cells look like healthy cells again. They employ cutting edge computer vision to extract the relevant features and score each compound. The general nature of this approach allows them to study dozens of diseases in parallel and has already yielded fruit, in terms of leads that are now in various stages of preclinical development.
The company is starting out in the rare disease space, where the unmet clinical need is well established (only 5% of the 2k known rare genetic diseases have an existing therapy). However, their platform is broadly applicable to dozens of therapeutic areas, including oncology, neuroscience, immunology, and infectious diseases. By introducing massive efficiencies into the drug discovery process, Recursion has the potential to dramatically reduce the cost of drug development and become a transformative leader in the pharma industry.
We are proud to be a part of Recursion’s Series B — and welcome to the CRV family!
— Dylan Morris
