Quick links: Breaking Election Invest Bitcoin Syria North Korea Startups Scandal
Wall Street Journal / Biz - Money

Glaxo’s Karenann Terrell Enlists AI in Understanding HIV Patients


Sara Castellanos

Karenann Terrell, chief digital and technology officer of GlaxoSmithKline plc, speaks Tuesday, Dec. 5, at the AI Summit in New York. Photo: Sara Castellanos / WSJ

NEW YORK – GlaxoSmithKline PLC is exploring whether artificial intelligence can be used to unravel the complexities of why certain HIV patients drop out of treatment, says Karenann Terrell, chief digital and technology officer.

The experiment is one of several projects using emerging technologies that are underway this year with the goal of saving patient lives and reducing the time it takes to get drugs to market, Ms. Terrell said in an interview with CIO Journal Tuesday at the AI Summit.

This year, the company partnered with startup Roam Analytics to analyze unstructured data such as doctor’s notes to look for clues as to why patients with human immunodeficiency virus drop out of treatment.

“Understanding why patients drop out of treatment is absolutely critical, and the only clues that exist are in unstructured types of data,” Ms. Terrell said.

Unstructured data can’t be stored in a conventional database, which limits the ways in which data such as doctor’s notes and pathology reports can be searched and analyzed. “With unstructured data, it’s very hard to find a signal,” she said.

Structuring the data requires advanced algorithms and natural language processing that can “read” the information in the documents and understand specific terms.

Ms. Terrell, formerly chief information officer at Wal-Mart Stores Inc., joined Glaxo in early September and is charged with bringing in new technologies for clinical trials and drug development. One of her goals is to use technology to significantly shorten the time it takes to get drugs to market, which is typically 12 years, she said.

In October, the company announced it joined a consortium aimed at using AI and data science to cut pre-clinical cancer drug discovery from six years to one.

Glaxo is contributing a dataset to the consortium that contains information about 2.1 million molecular compounds. The goal is to use supercomputers and artificial intelligence to mine Glaxo’s data and information from other organizations in the consortium, such as the National Cancer Institute, to better understand how certain molecules will behave in the body. The effort could ultimately help speed up the time it takes to identify successful compounds that could cure diseases, Ms. Terrell said.

This year, Glaxo has also begun experimenting with virtual reality to train employees in its advanced manufacturing facility in Ware, England, Ms. Terrell said.

In the future, she said, artificial intelligence combined with virtual reality could predict the next best action for individual employees, depending on their skill level. The goal is to ultimately reduce the time it takes to train workers and increase the time the machines are up and running, so that manufacturing doesn’t impede the time it takes for drugs to get in the hands of patients, she said.

Glaxo’s experiments are proof that technologies such as AI are now moving out of the consumer world and into the health-care sector, she said.

“Many of the problems with speed (to market) are going to require a brand new look at capabilities, and emerging technologies are incredibly important. A lot of that technology is new to pharma, but not new to the world,” she said.