Next-Gen Analytics Transforms Product Development

Developing new and improved pharmaceutical drugs is an expensive and time-consuming endeavor. Part of the challenge centers on researchers examining piles of medical literature— including studies and clinical research—in order to understand essential factors such as how illnesses are diagnosed and treated.

At Janssen Pharmaceuticals, a division of Johnson & Johnson, scientists and researchers are constantly looking for faster and more efficient ways to put data to work. As a result, the company—which produces drugs for an array of conditions, including hyperactivity disorder, infectious diseases, mental health problems and pain management—began looking for a technology solution.

In the past, “When we attempted to develop or improve drugs, it was necessary to read through hundreds and sometimes thousands of abstracts or full papers about outcomes from clinical trials,” says Soledad Cepeda, director of epidemiology at Janssen Pharmaceuticals. “We had to extract relevant information about subjects, conditions, treatments, how many people withdrew from the clinical trial and what adverse effects occurred.”

The data wound up getting slotted into an Excel spreadsheet and eventually appeared as fairly complex horizontal or vertical graphs. At that point, researchers conducted a detailed meta-analysis on all the information and graphs and drew conclusions about how to move forward with research and the allocation of resources. The entire process could take months and, in some cases more than a year.

“It was an extremely slow and painful process,” Cepeda explains. “Sometimes the information was outdated by the time we completed the process.”

Those days may be winding down. Janssen Pharmaceuticals has turned to IBM’s Watson analytics technology to make sense of scientific papers that detail outcomes of clinical trials. Watson reads and analyzes medical literature and extracts basic information that can be used for comparative analysis across a wide array of factors and variables. In addition to extracting data from the full text of studies, reports and other documents, the system also reads and analyzes graphs.

For example, if researchers are examining back pain and the use of a particular medication, they can focus on specific issues: how many participants in the clinical trial received a placebo, how many received the actual medicine, the number that withdrew due to adverse effects and how many withdrew because of a lack of efficacy.

By automating the process, Janssen Pharmaceuticals hopes to reduce tedious manual processes and generate more useful results. In the past, three people typically spent an average of 10 months collecting data and preparing it for analysis. Using Watson, the process can take place within minutes.

“Clinicians have the data at their fingertips, and they are able to use it to make better decisions,” Cepeda points out.

The system has the potential to radically transform the way the pharmaceutical firm approaches product development. “We are moving toward a far more sophisticated approach that yields better information about efficacy, side effects and many other factors,” Cepeda says. “We are able to answer questions and address challenges in a way that would have been impossible in the past.”