Sales of AI tools and drug development rights should fund the search for a drug to treat aging, says Alex Zhavoronkov
Sales of AI tools and drug development rights should fund the search for a drug to treat aging, says Alex Zhavoronkov
In an essay published on November 1, the founder of Insilico Medicine, an AI drug discovery company, explained why he is spending years of his own life pursuing this goal. He weighs the time he could spend earning money, traveling and resting against the additional years of life his work might give other people.
In 2013, Zhavoronkov published a book about the impact of biomedicine on the global economy. If a drug added a year to the life of each of the world's 8,3 billion people, humanity would gain 8,3 billion additional years of life. That potential gain explains his willingness to spend a decade of his own life trying to develop such a drug.
“The scale of the potential benefit justified the risk, even with a low chance of success.”
Zhavoronkov chose to build a commercial company to fund the years of development required. He discusses two other approaches: a large government program and a private laboratory with billions in funding secured at the outset. Insilico is intended to earn revenue by selling AI tools to pharmaceutical companies and licensing drug development rights, directing part of that revenue toward its long-term goal. When some programs are licensed early, a partner acquires the rights to a drug candidate before human trials and takes responsibility for further development, including those trials.
The author proposes pursuing a treatment for aging by starting with a specific disease, for which researchers can select patients, measure their response to treatment and seek drug approval. The search begins with proteins linked to both the disease and age-related dysfunction. Some AI models compare data on tissues, diseases and aging to propose these targets; others generate molecular structures designed to act on the selected protein. Chemists then synthesize the proposed molecules, and biologists test their effects.
One example of this approach is rentosertib, an experimental drug for idiopathic pulmonary fibrosis, a disease in which the lungs become scarred for an unknown reason, making breathing more difficult. Zhavoronkov explains that he chose this disease because it brings together processes associated with aging: tissue scarring and chronic inflammation. In a 12-week trial involving 71 participants, researchers assessed the drug's safety. To assess lung function, they asked participants to breathe in deeply and then exhale as forcefully and fully as possible, measuring the volume of air exhaled.
Blood samples were also collected from patients during the trial. Researchers later analyzed the protein profiles in samples from 42 of these participants using six models that relate protein profiles to age or mortality risk. This added model-derived measures of aging to the lung assessments in the same people.
AI speeds up target selection and molecule design. Synthesizing a compound, testing its safety, recruiting patients and monitoring their health take years, during which researchers' work must be funded. Zhavoronkov also proposes running many drug development programs in parallel so that, if one molecule fails, other potential drugs remain available for testing.