Skip to content

Report: Speed up drug development with artificial intelligence

But it says new legal, ethical, economic and social questions must be addressed

Senate HELP Chairman Lamar Alexander is among a group of lawmakers who requested the artificial intelligence report by the National Academy of Medicine and the Government Accountability Office. (Caroline Brehman/CQ Roll Call file photo)
Senate HELP Chairman Lamar Alexander is among a group of lawmakers who requested the artificial intelligence report by the National Academy of Medicine and the Government Accountability Office. (Caroline Brehman/CQ Roll Call file photo)

More and improved use of artificial intelligence, and an overhaul of medical education to include advances in machine learning, could cut down significantly the time it takes to develop and bring new drugs to market, according to a new joint report by the National Academy of Medicine and the Government Accountability Office.

Before that can happen, however, the United States must address legal and policy impediments that inhibit the collection and sharing of high-quality medical data among researchers, the report said.

“Machine learning holds tremendous potential in drug development,” according to the two-part report released Tuesday, which said such technologies could cut down the current time of about 10 to 15 years it takes to develop and bring a new drug to market. “In drug discovery, researchers are using [machine learning] to identify new drug targets, screen known compounds for new therapeutic applications, and design new drug candidates, among other applications.”

[Artificial intelligence is coming. Will Congress be ready?]

Researchers involved in drug discovery said infusion of machine learning technologies at the early stage of drug development could result in savings of between $300 million and $400 million per successful drug, the GAO said.

The first part of the report, prepared by the National Academy of Medicine, is titled “Artificial Intelligence in Healthcare: the Hope, the Hype, the Promise, the Peril.” The second part, prepared by the GAO, is called “Artificial Intelligence in Health Care: Benefits and Challenges of Machine Learning in Drug Development.”

The report was prepared by the two agencies at the request of a group of lawmakers including Tennessee Sen. Lamar Alexander, chairman of the Senate Committee on Health, Education, Labor and Pensions, as well as Republican Reps. Greg Walden of Oregon, Michael C. Burgess of Texas and Brett Guthrie of Kentucky.

[jwp-video n=”1″]

Annual U.S. health care spending is set to reach about $6 trillion by 2027, with the federal portion of spending on health care programs, which account for a quarter of the national costs, growing faster than the overall economy. The report found that application of artificial intelligence technologies could help identify new treatments, reduce failure rates in hospitals and result in more efficient and effective drug development.

But the application of new technologies also comes with new legal, ethical, economic and social questions that must be addressed, the report said.

Artificial intelligence-based tools can soon begin assisting patients with chronic conditions, the National Academy of Medicine said, describing some of the emerging applications of such technologies both for consumers and health care providers.

New devices could help patients with heart disease, diabetes or depression, and with taking their medications, modifying their diet, wound care and injections, the academy said.

Health experts also can use data drawn from wearable devices such as accelerometers, gyroscopes, microphones, cameras and smartphones for monitoring patients’ health and predicting risks, the academy said.

The report said startup companies with a focus on health and medicine have raised $4.3 billion to develop so-called smart clothing such as bras that can predict breast cancer risk and other clothes that can assess cardiac and lung conditions based on movement. In hospitals, clinicians are testing whether artificial intelligence technologies would allow them to personalize chemotherapy dosing — a kind of precision medicine that’s tailored to each patient.

Overhaul medical schools

To ensure that doctors and other health professionals keep pace with the infusion of new technologies, “medical education will need a substantial overhaul,” the academy said. Doctors and others need to be trained in data science and appropriate use of artificial intelligence products and services to the point where the new technologies “provide an assistive benefit to humans rather than replacing them.”

The artificial intelligence tools will “soon be essential to assist with the growing field of precision medicine,” the academy said, thanks to insights that algorithms can draw from a “stunning growth in the volume of information” drawn from routine medical procedures as well as from new products and biomedical research.

But researchers need to ensure that health data used to train artificial intelligence systems is representative of the population, the academy said. In the absence of such inclusive data, artificial intelligence tools could “exacerbate existing biases and inequities,” the report warned.

The GAO said that although the potential for the use of artificial intelligence technologies in cutting time and costs in drug development is high, some data challenges remain.

“Technological challenges include gaps in the underlying scientific data on mechanisms of disease, structure and behavior of complex molecules, and how to represent these data to algorithms,” the GAO found.

Researchers also said there was a shortage of high-quality unbiased data in part because of obstacles to “accessing and sharing data due to high costs and legal issues,” the GAO said.

The GAO recommended that policymakers should make several changes to federal law, including: focus efforts on funding basic research that can generate better data to improve the use of machine learning techniques in drug development; create mechanisms for sharing high-quality data among researchers; establish uniform standards for data collection and algorithms; and create a consistent message regarding machine learning in drug development.

Recent Stories

Strange things are afoot at the Capitol

Photos of the week ending May 24, 2024

Getting down on the Senate floor — Congressional Hits and Misses

US-China tech race will determine values that shape the future

What’s at stake in Texas runoff elections on Tuesday

Democrats decry ‘very, very harmful’ riders in Legislative Branch bill