The average time to market for a new drug is around 12 years, at a cost averaging $2.6 billion, according to 2014 research by Tufts University. And there’s no indication that either the cost or time involved has gone down since then.
Few drugs make it from phase one clinical trials to pharmacy shelves.
Artificial intelligence (AI) shows tremendous promise in reducing both cost and time to market because of important technologies that fall under the AI umbrella like:
- Extracting meaning from “unstructured” data like scientific papers and physician notes
- Analyzing massive quantities of data to derive previously hidden drug-disease correlations
- Helping identify compounds that are likelier to make it from Phase one testing to market
Pharmaceutical AI today is tackling management of existing data, bringing AI and R&D in-house, finding new investors, and studying repurposing of past drugs that didn’t make it to market.
Lack of Data Is Not a Problem
AI technologies have plenty of data to work with, for the most part, and more is being created with each passing day. The issues with data used by AI technologies have more to do with ensuring data quality and finding ways to share existing data. To this end, data gatherers are focusing more energy on data collection processes, to help ensure high quality from the outset.
One potential problem is a tendency by pharmaceutical companies to hoard data. After all, data is extremely valuable and was obtained at high cost. Finding new methods of knowledge sharing will be important to turn the conversation away from amassing data and toward sharing it. Negotiating with data owners is likely to become increasingly important in coming years in order to determine who can do what with shared data.
Startups Want to Bring AI, R&D In-House
Many of the startups that have emerged at the intersection of AI and pharmaceutical research are prioritizing gathering of huge amounts of existing data in order to identify new disease targets and develop new drugs. Bringing both AI and R&D in-house is also a priority of many pharma AI startups. With AI reducing R&D costs, this may be possible for them.
One issue startups and big pharma alike must grapple with is the patentability of AI and the innovations derived from it. Traditionally, patents are used to fund future R&D and protect innovations, and it’s as yet unclear what effects AI will have on meeting the inventive step requirements for obtaining patents.
Technology outpaces legal precedence, so it may take time for legal consequences of AI in pharmaceuticals to become apparent.
Big Pharma More Cautious
Big pharma firms are moving more slowly into pharmaceutical AI, though they are using it. Pfizer, for example, has many AI projects underway, but most of them are not related to core drug development. Furthermore, big pharma is not providing the dollars driving the pharmaceutical AI ecosystem. Rather, investors are funding it.
This is not to say that AI startups are going it alone. Many partnerships and new research “ecosystems” are being developed every day. Some startups complain that it’s too difficult to do deals with big pharma, and other partners are appearing in their place. A company in China called WuXi Apptec is one of the strongest emerging partners for startups engaged in early-stage drug development.
Rare Diseases and Drug Repurposing
Semantic technologies and unstructured big data are expected to identify many times more patients with rare diseases than traditional research methods have. History shows that the most severe diseases result in data that is largely unstructured, including first-hand data from patients, test results, and physician notes.
Historically, drug companies that miss endpoints, or for whom hypotheses fail, abandon drugs and move on to the next product. But some of those abandoned compounds have potential for uses other than those originally intended. With AI that can identify population subsets for whom abandoned, existing drugs will work, drug repurposing stands to make pharmaceuticals a far less wasteful industry.
Clearly, AI is part of the here-and-now of drug development and not some vague concept on the horizon. Disruption is already occurring, and there’s not yet a clear picture of which technologies will prove most valuable. The many startups focusing on pharmaceutical AI are likely to coalesce some as new partnerships are formed, and as big pharma companies see the value in acquiring some of them.
Drug development companies will have to bet on technologies without knowing outcomes, but pharmaceutical companies have done that for a long time already. Investors in pharmaceuticals must follow new developments in AI as well as the many ways today’s pharmaceutical AI startups are putting the technology to work.
