Building a virtual model to solve human aging

Peter Fedichev explains how Gero is using AI and activity data to accelerate our understanding of human aging.

We recently reported on AI firm Gero’s involvement in a new study that demonstrated the ability to quantify the link between resilience and aging. And, when we first spoke with the Singapore-based company’s co-founder Maxim Kholin a year ago, we learned how the company was leveraging theoretical physics and AI to find new drug targets for aging from analysis of data from biobanks.

Longevity.Technology: The development of its GeroSense AI app and the new paper demonstrates Gero’s focus on leveraging AI across a diverse range of areas to solve the mysteries of aging. But how do these different elements come together as part of the company’s overall strategy? We spoke with co-founder and CEO Peter Fedichev to find out more.

Fedichev explains that Gero was built around bringing the AI data analysis of aging in humans to a level comparable to current tools for weather prediction or financial market analysis.

“We aim to combine mathematics and physical sciences, using dynamical systems theory, to the point that we can predict the future from the current state of an individual,” he says. “A virtual human model, which is combined with molecular data like genetics, could be transformed into a generator of experimental ideas on how to control aging in humans.”

Peter Fedichev and the team at Gero.

What Gero is doing by using GeroSense AI to measure aging based on activity levels could, in some ways, be compared to the development of biological clocks of aging. But Fedichev points out a key distinction.

“What we do differently is that we focus on longitudinal analysis,” he says, explaining that biological age measurements tend to fluctuate a lot depending on lifestyle variations, such as how well you sleep on a given night. “Our data suggests that this is not a biomarker of age, but rather a biomarker of stress – and aging is a stress, which is unrecognised and unrepaired by your body.”

Resilience: a biomarker of aging

Gero’s interest in stress and how well we recover from it is central to the recent paper in Nature Communications. In particular, it focuses on our ability to bounce back from various stresses as we age.

“This phenomenon is called resilience, and it can be measured directly from longitudinal tracking of biological age,” says Fedichev. “From the data, we can see how quickly you go back to the norm – we can actually measure it, it’s a quantitative test. So you can actually calculate the recovery rate from the data, and which I believe is the most important biomarker of your aging.”

In order to measure our ability to recover, Gero had to collect a lot of longitudinal data, which it was able to gather from users of GeroSense AI, its free app for tracking and measuring biological age.

“Our users donate data to us in such a way that we see fractions, components of an individual’s aging trajectory,” says Fedichev. “Every iPhone user these days has maybe three to five years of physical activity measurements, which are individual trajectories for biological age. So we can see how resilience changes over time.”

From this data, Gero was able to observe that the number of people with low resilience exponentially increases with age. While this may seem obvious, Fedichev says it had not previously been possible to measure or quantify this change in resilience.

“These are very small changes, that can only be seen in the trajectories you can see in longitudinal data,” he says. “So we believe that resilience is a novel biomarker that highlights the vulnerable sub-population of people who are not able to recover, even from reasonable stresses. And it doesn’t depend on age – if your resilience is lost at any age, then you are vulnerable.”

Resilience and radical life extension

Fedichev believes that the newfound ability to track and measure resilience holds great potential for the development of interventions that target aging.

“Resilience tells us about maximum lifespan, whereas biological age is a biomarker that is only important for the last 10 years of your life, when you’re already unstable,” he says. “The interventions that are currently studied in mice target biological age, and I believe those interventions will most probably only increase lifespan past health span. So a drug that affects only biological age without affecting resilience would produce only an incremental effect on lifespan.”

While he acknowledges that it would still be “cool” if a drug were be able to achieve an incremental increase of 15 years or so, Fedichev believes that targeting resilience has a much greater potential.

“If we want to go for radical life extension, you should not concern yourself with drugs that only look at biological age – you should be specifically looking for drugs that improve resilience,” he says. “I think what is also very important is that if a drug can improve resilience, it would also improve performance, which means that you could possibly cope with a much higher level of stress.”

Uniting the data

Fedichev estimates that if GeroSense is able to collect the data for one million users for one year, then the company will be able to “solve resilience”. On top of the resilience data it is collecting via GeroSense, he says that the additional work Gero is doing through AI analysis of biobank data will help the company get closer to the discovery of aging interventions with real potential.

“Using the biobank data, we can do genetic studies to find the genes that are responsible for resilience, and, most importantly, to identify drugs already on the market that may affect resilience. We still have a lot of work to do – but we have the tools to do it.”

Images courtesy of Gero