By Deborah Borfitz
July 17, 2020 | Newly launched Deep Longevity, a spinoff of Insilico Medicine, will be using artificial intelligence (AI) AI to tackle age-related diseases using a constellation of “aging clocks” as yardsticks of health and the effectiveness of lifestyle interventions and dietary supplements—as well as develop new aging clocks in collaboration with concierge longevity practice Human Longevity, Inc. (HLI). The clocks could also serve as auxiliary biomarkers in clinical trials of geroprotectors, a recently identified class of anti-aging compounds that include the supplement nicotinamide riboside, the commonly prescribed diabetic drug metformin, and the anti-cancer drug rapamycin.
So says Alex Zhavoronkov, Ph.D., founder and CEO of Deep Longevity and patent holder on biomarkers used in some of the aging clocks. Zhavoronkov has been conducting research in the emerging field of longevity medicine since 2006 and eight years ago created Insilico Medicine to combine genomics, big data analysis, and deep learning for in silico drug discovery.
In the scope of collaboration with HLI, Deep Longevity will develop and provide customized predictors of human biological age to clinicians at Health Nucleus, the brick-and-mortar health wellness space of HLI located in San Diego, California. Its clientele is among the “ultra-rich” who demand and receive the best available care in the land, and that makes them ideal first adopters of the technology before the price tag falls within reach of most consumers, says Zhavoronkov.
HLI clients will have exclusive access to the customized AgeMetrics reports issued by Deep Longevity that provide a snapshot of their biological age and some of the contributing factors, he continues. The aging clocks will initially include AI predictors trained on clinical blood tests as well as those based on an individual’s behavior, photo, medical history, microbiome, transcriptome, methylation, and heart rate variability.
The clocks in some cases correlate with a disease, such as diabetes and sarcopenia, but the qualifying criteria is that what they measure associates with mortality and perhaps also a clinical intervention, says Zhavoronkov. The interventions will initially target lifestyle habits such as diet, sleep, exercise, and supplements—to the extent they prove impactful—but could later include drug therapies and nonpharmaceutical treatments, some of which the aging clocks might help bring to market.
Each clock generates a mathematical score (aka biological age) and collectively they produce a weighted average of all those individual scores referred to as AgeMetric, Zhavoronkov says. Clinicians can readily see which aging clocks are having the most impact on the aging process of a specific patient.
The AgeMetric system works much like the one developed by Digital Science (Altmetric Explorer) to measure the popularity of research papers based on factors such as citations in other journals and the number of retweets and Facebook shares. The overall aging score can be visualized as a “beautiful flower” to which each data type adds a new leaf, as Zhavoronkov describes it.
Aging clocks based on a recent blood test, behavioral survey or photo could be factored into the predictive score right away and then get adjusted when clocks requiring sophisticated laboratory testing, such as transcriptomics and microbiomics, can be added to the calculation, he says. Deep Longevity stores the scores but none of the personal medical data used by the AI algorithm.
Repeat testing with the aging clocks can happen at intervals deemed best by a patient's physician, Zhavoronkov says, although it is “usually a good idea with every medical checkup to see how those clocks change over time with your health status.” This would be particularly important for any aging clocks that prove to be disease-relevant for a patient.
“Very few physicians have the advanced knowledge of aging clocks, a field pioneered by a UCLA scientist Steven Horvath in 2013,” says Zhavoronkov. “AI biomarkers of aging are even newer. It is an emerging field that did not exist just a decade ago and the research publications are coming out almost every month. Deep Longevity will help accelerate the deployment of these aging clocks into the clinic.”
The AI-powered aging clocks are backed by science, he says, unlike many of the anti-aging remedies currently being peddled. By his estimation, at least 90% of the longevity market comprises fraudulent products that don’t work. Supplements and anti-aging clinics alone generate upwards of $150 billion per year globally, which is over 10% of the global pharmaceutical market that was valued at $1.25 trillion in 2019.
Deep Longevity’s aging clocks can predict an individual’s biological age in a more comprehensive way than conventional methods that are predominantly based on methylation levels, Zhavoronkov says. Methylation status, while highly predictive of someone’s age, is not routinely used in clinical practice because it is hard to tie those values back to the causative underlying processes so they can be addressed. “Simple data types like behavioral data, imaging data, lab tests, and even microbiome data provide for easier starting points.”
Knowing researchers have had limited success developing other types of aging clocks, scientists at Deep Longevity decided to “go broad” and incorporate clocks based on regular blood tests that physicians would immediately understand how to modify. It also uses an algorithm indicating when biomarkers within the normal acceptable range can be “tweaked a bit” to reduce an individual’s biological age. “It’s very interpretable and very actionable.”
Deep Longevity will also be developing a Young.ai app that will provide clients of elite longevity and preventative care centers like Health Nucleus with daily monitoring of their aging rates and help keep them engaged in the continual tracking of their health, says Zhavoronkov. Exploding popularity of consumer wearables and health apps has created a cultural phenomenon, referred to as the “quantified self,” which should provide Deep Longevity with a ready market for the age prediction app when it gets released to the general public later this year.
The field of longevity medicine has also been gaining credibility over the past few years, Zhavoronkov says, thanks to people like Jim Mellon, co-author of the book Juvenescence and one of the founders of a company by the same name. Other industry drivers spurring credible companies in the field include Peter Diamandis, a Harvard- and MIT- trained physician and entrepreneur; David Sinclair, a professor and serial entrepreneur out of Harvard; and Wei-Wu He, a Baylor- and Harvard MGH-trained scientist who is now the chairman of Human Longevity.
Parallel Ecosystems
As announced on July 14, the startup has attracted funding from prominent investors around the globe, led by ETP Ventures and Human Longevity and Performance Impact Venture Fund (“HLPIVF”). Other named investors include BOLD Capital Partners, Longevity Vision Fund, Formic Ventures, and LongeVC; undisclosed investors include those with celebrity status and prominence in the U.S. biotechnology sector.
Zhavoronkov adds that management at Insilico Medicine, including senior research scientist Polina Mamoshina, also participated in the financing round. “We’re all very confident this venture is going to be very big.”
The entire collection of aging clocks sits in an Amazon cloud, allowing Deep Longevity to create a “longevity as a service” product suite to help people live longer and function better, he says. HLI is separately growing an ecosystem of startups around its genomic-powered, precision medicine business model that includes the planting of new Health Nucleus locations around the globe.
A parallel in the fine dining world for the relationship between Deep Longevity and HLI is that of a gourmet mustard sauce or specialty beer doing its big book of business with a Michelin 3-star restaurant, Zhavoronkov says.
Age-Related Matters
Outside of its collaboration with HLI, Deep Longevity has conducted research concluding that COVID-19 is a “gerolavic” disease, meaning “harmful to the elderly,” says Zhavoronkov. In April, in a paper published in Aging (DOI: 10.18632/aging.102988), he points to statistics from the pandemic indicating the majority of the infected population is 50 or older and almost everyone who dies of it is 60-plus years of age. “When a young person dies it makes the news.”
His hypothesis, now being tested with the aging clocks, is that people who are biologically (but not necessarily chronologically) older are more likely to be infected with COVID-19 and have a more severe course of the disease. Blood samples have been collected from participants around the world, and more are needed. “Unfortunately, it’s not easy to share this kind of data because there’s no central repository for blood tests,” Zhavoronkov says, “but we’ve seen some signals.”
As he notes in the Aging article, geroprotectors that help reverse some aging clocks may be protective against COVID-19. Some researchers are already acting on the suggestion, including a trial of erythromycin that recently launched at the University of Cincinnati, says Zhavoronkov.
Moving forward, he expects the aging clocks will be used more broadly in clinical trials as auxiliary biomarkers to help study sponsors learn more about where an intervention is having an effect, and at levels that might be missed by traditional statistical analysis tools.
Deep Longevity also plans to find a customer base among life insurance companies whose actuarial models would be more accurate if they included the biological age of people, Zhavoronkov says. Innovations in the space have been limited to common behavioral modifications and broadening of exclusion criteria.