Towards Ionomic Aging Clocks
The ionome is the elemental composition of a tissue, organ, or individual. This composition changes over the course of aging, and may do so in ways that allow the production of an aging clock, a measure of chronological or physiological age. This line of development adds to work on the well-known epigenetic clocks, proteomic clocks, and other assessments of age constructed from algorithmic compositions of simple biomarkers. At the end of the day, all of these approaches need a great deal more validation if they are to be used as originally intended, as a way to rapidly assess potential rejuvenation therapies and thus speed up the field. Since it remains quite unclear as to what exactly these clocks measure, meaning which processes of aging cause the clock numbers to change, the results are not yet actionable.
Aging involves coordinated yet distinct changes in organs and systems throughout life, including changes in essential trace elements. However, how aging affects tissue element composition (ionome) and how these changes lead to dysfunction and disease remain unclear. Here, we quantified changes in the ionome across eight organs and 16 age groups of mice. This global profiling revealed novel interactions between elements at the level of tissue, age, and diet, and allowed us to achieve a broader, organismal view of the aging process. We found that while the entire ionome steadily transitions along the young-to-old trajectory, individual organs are characterized by distinct element changes.
The ionome of mice on calorie restriction (CR) moved along a similar but shifted trajectory, pointing that at the organismal level this dietary regimen changes metabolism in order to slow down aging. However, in some tissues CR mimicked a younger state of control mice. Even though some elements changed with age differently in different tissues, in general aging was characterized by the reduced levels of elements as well as their increased variance. The dataset we prepared also allowed to develop organ-specific, ionome-based markers of aging that could help monitor the rate of aging. In some tissues, these markers reported the lifespan-extending effect of CR. These aging biomarkers have the potential to become an accessible tool to test the age-modulating effects of interventions.
If aging is caused by cellular damage, senescent cells, dna damage in oncogenes leading to cancerous cells, cell loss, crosslinks, mitochondrial damage, waste build up inside lysosomes and outside cells through amyloidosis etc then the only true aging clock is one that measures this damage and puts it into a model that maps these input measures to the output age. If (IF) we could measure this in enough people AI could be used to derive the model/weights. Any other aging clock would need to be a reliable proxy to indirectly measure this damage. I doubt the ionome would be a good proxy for these damage types?
Also if we could measure the damage and apply AI to derive the model, then the weights in the neural network model might tell us something about which damage causes the biggest aging contribution or earliest aging contributions? Aubrey recently said more work was needed to quantify this. I would have thought this would be valuable in funding research and in treating people depending on their exact age and damage profile to achieve longevity escape velocity. (Recognising of course that all forms of damage would need to be treated at in least in part in an chronologically old individual to even hope for lev).
One more thought, AI could also be used to map from measured damage to individual age related diseases showing which damage types promote which disease. So if we can measure all the forms of damage in a big cohort of people with a given age related disease we could maybe train the AI to start to predict which individuals without the disease might expect to be on track to get it. This would be extremely powerful evidence proving further the SENS model and in combination with the AI aging clocks described in the comments above could powerfully support that age related disease and damage accumulated from age are one and the same. Before anyone says, then yes we would need a massive amount of reliable measurement data to make this work and/or better artificial intelligence technology than we have today (so neither of which we can do today).
@David Luck
we currently have a very bad and inexact definition of age. The SENS view is that there is no difference between disease and aging. Some indicataor (like tree rings) might show 500 years but if there is no real damage what does it mean ?
I am skeptical about AI helping here. It not very hhmmm intelligent. Just a fancy engine to do find and apply statistical classifications. The current beleading edge needs enormous amounts of data. If there was an easy and non-invasive way to scan each cell and tissue that would provide the raw data. For now, the data is also quite limited. And for limited data, we puny meat-bags, are much better at interpretation and classification.
Hi Cuberat, AI would absolutely help lets avoid the discussion about how intelligent it is for the moment, and lets consider that AI powered Go trained from scratch by playing itself over and over very very fast in 3 hours how to beat all humans on the planet and there are countless other examples, it is not applying statistical classifications in mathematical terms it is converging on the correct high dimensional mathematical function to map one very large dataset to another. We puny meat bags are definitely not better than it at classification it beats us already, although we are better at common sense knowledge, learning from one example etc, etc. It would definitely help with our current bad and inexact definition of age and that is my whole point, although I also said it cant do it yet, we should work on this!