Digital Data in Health Care Promises Much, Has Limits

Laurie Glimcher, CEO of the Dana-Farber Cancer Institute, described the use of AI in treating cancer.

Technology is reshaping health care, from pharmaceutical research to detecting opioid addiction, even though the pace of change isn't as dramatic as some had hoped a few years ago, industry leaders said at The Wall Street Journal Health Forum on Tuesday.
 Novartis AG Chief Executive Vas Narasimhan said opportunities presented by artificial intelligence are, for now, on the margin. “It's another tool in the toolbox,” he said. Doctors, insurers and drugmakers are experimenting with digital technology at a brisk pace after years of watching internet access and smartphones disrupt nearly every other economic sector.
 They hope automation and data-driven predictions can create new efficiencies in their industry, which contributes nearly 18% of U.S. gross domestic product.
 Pharmaceutical companies hope artificial-intelligence algorithms can help scientists discover new drugs and improve the industry's abysmal success rate in bringing treatments to regulatory approval.
 Novartis has invested heavily in AI programs, which aim to make progressively more accurate predictions over time, in hopes of squeezing more products out of its nearly $9 billion annual research- and-development budget.
 So far, though, the Swiss drugmaker has had the most success using artificial intelligence to forecast financial results and clinical-trial enrollment, Dr. Narasimhan said at the WSJ forum in Washington. Dr. Narasimhan said he is skeptical that AI can keep up with rapid advances in complex diseases like cancer, or that it can be better than humans at predicting which drugs will work.
 Pfizer Executive Chairman Ian C. Read agreed, saying in a later session that AI's capabilities are restricted by gaps in human knowledge about how diseases function in the body. “Using AI in drug discovery is extremely difficult and unlikely to be productive in the near term because our understanding of biology is not as deep as we'd like to believe it to be,” Mr. Read said.
 Still, even applying digital record-keeping and communication could add up to significant financial savings, both men said. Dr. Narasimhan said studies show drug companies could cut about 20% from research- and-development costs by using certain technologies at a broad scale.
 Mr. Read said a big contributor to high development costs is the antiquated system of running clinical trials relying on paper records and in person visits with trial sites.
 Dana-Farber Cancer Institute CEO Laurie Glimcher was more optimistic about AI. She said researchers at the Boston- based medical center are exploring the use of machine learning to better analyze tumor samples from biopsies. Dana-Farber offers to sequence the DNA of all its patients and combines the data with patients' medical records and treatment outcomes.
 That type of analysis could help doctors within a few years predict the one or two drugs most likely to help a patient based on their genetic profile and medical history, Dr. Glimcher said. The large volumes of data required to do the analyses will require sophisticated computer algorithms, she said.
 “We want to put our patients on the right drug at the beginning. Because they don't have time to wait,” Dr. Glimcher said.
 Centene Corp., a St. Louis based health insurer, has developed algorithms to predict which patients are likely to develop opioid addiction or which are already addicted, and then work with the patients' doctors on a recovery plan, said Chief Executive Michael Neidorff. Still, insurers have to be mindful that they don't allow predictive analytics to trump physician judgment, he said.
 “You have to understand what data's limitations are,” Mr. Neidorff said.

 By Joseph Walker, Peter Loftus and Brianna Abbott  


Add new comment