IBM to wind down Watson's work in AI-based drug discovery: report

IBM is ending new sales its Watson programs aimed at artificial intelligence-based drug discovery and development due to poor financial returns, according to a report from Stat.

The news follows Big Blue’s decisions last year to move away from offering Watson in hospital management settings, not to mention the departure of the general manager of the division’s health-focused unit last October.

IBM has previously rolled out some high-profile partnerships, though not entirely in drug discovery, including with Pfizer, Novartis and Illumina in cancer research, as well as with Teva to explore drug repurposing. However, even with these deals, sources told Stat the initiative was still not bringing in enough money.

According to those sources, IBM employees were recently notified of the decision and have begun reassessing Watson’s prospects in the broader biopharma industry. More officially, the company said it will continue working with drugmakers who currently employ Watson for their R&D work.

“We are focusing our resources within Watson Health to double down on the adjacent field of clinical development where we see an even greater market need for our data and AI capabilities,” an IBM spokesman told Stat in a statement.

Outside of drug and clinical development, Watson Health recently inked a $50 million, 10-year investment into a collaboration with Brigham and Women’s Hospital and Vanderbilt University Medical Center to research public health issues and patient safety.

RELATED: IBM Watson Health chief Deborah DiSanzo steps down from post

But over the past few years, Watson has been no stranger to stumbles. A 2017 investigation by Stat found that IBM’s Watson for Oncology initiative—which aimed to use AI and natural language processing of medical records to help recommend personalized courses of cancer treatment—was a long way from being able to meet its promises.

Though pitched by marketers as a way to generate new insight and approaches to care, the system has been beset by obstacles in interoperability and data collection—namely the labor-intensive work required by physicians and others to keep the program manually updated with treatment recommendations and data for only a handful of different cancers.