The exuberance is all the more notable because venture capital investment in other sectors has slowed in recent years. “It’s very much a tale of two cities,” said Patrick Malone, a principal at KdT Ventures focused on life science startups. While traditional drug development companies are seeing a colder VC market compared to two or three years ago, he said, an emerging field of AI-backed therapeutics startups is drawing capital hand over fist.
The stampede isn’t limited to biotechnology. Companies are applying machine learning to virtually every aspect of healthcare, from pathology to billing. Healthcare as a sector has remained stubbornly manual even as automation in manufacturing and simple computing powered many of the biggest industries of the last century, said J.C. Lopez, a principal on the healthcare investment team at New Enterprise Associates.
That’s true for both biotechnology and the mundane back-office functions of providers, but for different reasons. In the case of drug discovery, identifying the molecules that affect diseases is like finding a needle in a haystack, involving massive amounts of trial and error, time and money. On the administrative side, the variability of human health and the complexity of medical billing that accompanies it has been a morass too thick for traditional computing to crack.
Not a Modern Healthcare subscriber? Sign up today.
That is still more or less true in the case of drug development. “You do still need to have that human component,” said Lopez, who has a medical degree. But some of the ever-increasing data being collected in the lab can now be analyzed more efficiently, “which is what AI is good at.”
Hype around the power to analyze vast sums of data using large language models and other types of machine learning has seeped into almost every sector.
Its turning point in healthcare may have been in 2021 when Google unveiled the results of AlphaFold, a software using huge datasets to predict the shape of proteins, a key function in the discovery of new pharmaceuticals. That year saw an explosion of venture capital funding in AI-backed health startups, with $1.9 billion invested in New York companies, close to quadruple the years before and after, according to PitchBook.
In September, Chai Discovery, a drug discovery startup that began in offices in New York before moving to San Francisco, raised $30 million in seed funding led by Nolita-based Thrive Capital. Chai is part of a subset of startups marketing predictive software for therapeutics companies to apply, rather than developing drugs itself. The sector is hoping to make it cheaper and faster to bring a drug through clinical trials to market, which currently takes around a decade and $2 billion on average.
Download Modern Healthcare’s app to stay informed when industry news breaks.
Some investors are sober about the limits of AI’s power to spur a pharmaceutical revolution. Predicting protein shape is just one of hundreds of stages in the drug discovery process. Bringing those components together in a model that can spit out a new drug at the push of a button has remained beyond the reach of AI.
Investors have a “tendency to overgeneralize success stories,” said Malone, who also has a medical degree. “It’s impactful but it’s not going to increase things by an order of magnitude, which I think if you take some of the interest in this space at face value, people may be expecting.”
The biggest cash cows are the companies applying advanced computing to the administrative functions of a doctor’s office, Malone said, which text-based large language models are particularly adept at tackling.
That is certainly bearing out in New York. Nirvana, a Lower Manhattan-based startup that aims to use AI to process insurance claims, raised $24 million in a Series A in September. Another company, Flatiron-based Spring Health, which claims to use AI to improve patient referrals, closed a $100 million Series E in August.
If AI becomes able to automate some of the documentation providers must produce, “it’s very easy to imagine how that becomes a massive, massive market,” Malone said.