GAO, NAM: AI and ML Could Speed Drug Development
- By John K. Waters
Artificial intelligence (AI) and machine learning (ML) technologies have enormous potential to enable a much more efficient and effective drug development process. That's according to a jointly published report ("Artificial Intelligence in Health Care: Benefits and Challenges of Machine Learning in Drug Development") by the General Accountability Office (GAO) and the National Academy of Medicine (NAM).
The GAO, the U.S. agency that tracks how the government uses taxpayer dollars, and NAM, an American nonprofit organization that provides national and international advice on health, medicine, health policy and biomedical science, concluded in the report that AI/ML tech is already having an impact on the current drug development process, which historically has been lengthy and expensive, taking 10 to 15 years to develop a new drug and bring it to market. These technologies have the potential to expedite the discovery, design and testing of drug candidates, decreasing the time and cost required.
"These improvements could save lives and reduce suffering by getting drugs to patients in need more quickly," the report's authors wrote.
The potential of ML in particular is being demonstrated in application to the first three steps of the drug development process: drug discovery, preclinical research and clinical trials.
"In drug discovery, researchers are using ML to identify new drug targets, screen known compounds for new therapeutic applications, and design new drug candidates, among other applications," the report states. "In preclinical research, ML can augment preclinical testing of drug candidates and predict toxicity before human testing. Researchers are also beginning to use ML to improve clinical trial design, a point where many drug candidates fail. These efforts include applying ML to patient selection and recruitment, and to identify patient populations who may react better to certain drugs, thus advancing towards the promise of precision medicine."
Perhaps the key behind the growing efficacy of AI/ML technologies in modern medicine is the "stunning growth in the volume of information generated through routine health-related processes and from products of health, healthcare, and biomedical science research," the report observed. "[T]he ability to glean insights from the enormous body of data points generated daily from mobile apps (m-Health) and sensors will require the capacity for simultaneous data processing from multiple sources. The increasing availability of environmental and geospatial sensors developed on digital platforms contribute yet additional data universes requiring AI/ML before incorporation into predictive modeling tools."
But the report also warns that growing use of AI applications in health care decision-making increases the need for transparency in algorithms and data sources, keeping in mind that "the need for algorithmic transparency is context-dependent, based on risk and intended use." As an example, the report offers "a high impact AI tool with immediate clinical implications, which warrants more stringent explanation requirements than a tool with a proven record of accuracy that is low risk and clearly conveys its recommendations to the end user."
"Therefore, AI developers, implementers, users, and regulators should collaboratively define guidelines for clarifying the level of transparency needed across a spectrum," the report's authors advise.
Another challenge the report highlights: adapting medical education to an AI/ML-enabled health services environment.
"Given the necessary reliance on IT and ML to help health professionals keep pace with the rapidly growing knowledge base," the report reads, "medical education will need a substantial overhaul. This needs to happen with an added focus on the use of AI as a routine decision-assistance tool. Training programs across multiple professions will require a focus on data science and the appropriate use of AI products and services. The bridging function to patient and consumer comfort levels with these emerging technologies will also need to be established to secure the bond of confidence between clinicians and their patients. Ultimately, the goal is to build competency in AI and data science to the point that health care AI provides an assistive benefit to humans rather than replacing them. For this reason, the near-term focus might be better termed 'augmented intelligence.'"
The full report is available online.
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at firstname.lastname@example.org.