How AI Is Transforming Drug Creation

How AI Is Transforming Drug Creation

On a recent Friday in Boston, Randell Sanders gave a nurse two samples of his blood, plus a sample of urine and saliva. Clinicians would test some of the samples to see how he is responding to treatment for pancreatic cancer.

But samples also were sent to a lab where computers using artificial intelligence are changing the way pharmaceutical companies develop drugs.

The idea is that machines, which are adept at pattern recognition, can sift through vast amounts of new and existing genetic, metabolic and clinical information to unravel the complex biological networks that underpin diseases. That, in turn, can help identify medications likely to work in specific patient populations, while simultaneously steering companies away from drugs that are likely to fail.

In the past, drug companies have used artificial intelligence to examine chemistry — whether a drug might bind to a particular protein, for instance. But now the trend is to use AI to probe biological systems to get clues about how a drug might affect a patient’s cells or tissues.

Biological insights driven by machine learning also could help pharmaceutical companies better identify and recruit patients for clinical trials of therapies most likely to work for them, perhaps boosting the chances of those medications’ getting approved by regulatory agencies such as the Food and Drug Administration.

Data from the samples produced by Mr. Sanders, 64, a U.S. Navy veteran, will become part of the database in a $17 million, seven-year study known as Project Survival, bankrolled by Berg, a Framingham, Mass., biotech firm that is one of several companies in the U.S. and Europe using AI to make drug research and development less expensive and more efficient. Mr. Sanders says he agreed to take part in the study in hopes that it might “help the next person.” Intelligent machines will scour his samples and genes, along with those of hundreds of other patients, for molecular fingerprints, or biomarkers, that could later be used to help measure a specific drug’s impact and to identify patients in which such a drug is likely to be most useful.

The big difference between AI-driven drug trials and traditional ones, says Niven Narain, chief executive of Berg, is “we’re not making any hypotheses up front. We’re not allowing [human] hypotheses to generate data. We’re using the patient-derived data to generate hypotheses.”

Project Survival is part of a larger research program to develop therapies with the help of intelligent machines. Other efforts to leverage AI technology in pharmaceutical research include using it to find new drugs or new uses for already approved medications, as well as speeding up clinical trials by improving patient recruitment and site selection, according to a May 2017 report by analyst Datamonitor Healthcare.

Some companies, such as Numerate Inc. in San Bruno, Calif., and BenevolentAI Ltd. in London, are developing their own molecules and licensing them to drug-industry clients. Others, such as International Business Machines Corp. , Atomwise Inc. in San Francisco and Insilico Medicine Inc. in Baltimore, are forming research partnerships with universities and nonprofits or setting up AI services aimed at drug companies.

For example, Merck & Co. is using Atomwise’s deep-learning technology to identify compounds that could be developed into medications for neurological conditions, according to David Rosen, an associate principal scientist at Merck Research Labs in Palo Alto, Calif. Recently, there’s been growing interest in leveraging this type of AI for health-care applications, in part due to the vast improvements deep learning has enabled in applications like machine translation and computer vision, which also rely on pattern recognition.

In January, GlaxoSmithKline PLC and Lawrence Livermore National Laboratory in Livermore, Calif., announced a partnership to use AI for pharmaceutical R&D.

Posted on 7wData.be.