I am happy to comment on this. I have previously used two labs at the same time and indeed there can be significant variations. Some are clear lab errors (such as delays in testing for creatinine). Others can be methodology issues. I, however, intend calculating the variation and using that as an adjustment for values so that historically I can compare things more accurately.
I got a really nice MCV (for me) last week (93). This week is not so nice (96.1), but a different lab. Sadly I cannot use these two labs simultaneously. One is something I go to a particular pharmacy and they send off the blood. The other I take the blood sample to the lab.
My view is that values from the same lab are probably comparable.
Seems Aging.Ai updated their algorithms. My age has jumped from 29 to 35 on the same data taken a month ago. I am unsure why the age changed although my biomarkers for blood glucose and LDL went in the wrong direction.
However, I checked my legacy score from early 2022, and it still shows 28. So, something I have done has caused my aging to accelerate. I guess I should reduce my Rapamycin dosage down to 6 mg equivalent as that was the biggest change I have made from then until now. I was not yet taking Rapa at that time.
The Levine calculator shows an age of 36, which is in-line with the results from 2022. So, whatever I changed affected Aging.AI but not Levine.
Rapamycin use did bring down my epigenetic age by 7 years on an epigenetic age spit test.
Itās so confusing that taking Rapamycin and other treatments will make you older on one calculator, the same age on a second calculator, and younger on a third!
This is useful confirmation of what should be expected from such processes. It is frustrating but since the tools and calculators are very rough, Iām trying to think of it like āsignificant digitsā in my old math class.
I wonder what @ConquerAging experience has been with random test result fluctuations impacting his analysis? He certainly has done a lot of tests and a lot of analysis.
Many months ago Aging.ai gave a much more flattering age reduction than the Levine-based spreadsheet. Now Aging.ai gives me almost identical results as the Levine-base spreadsheet.
They claim that the algorithm is constantly learning based on the inputs from users.
These are based on numbers from my most recent blood test. Collected: 08/30/2023 09:06
I am happy with the results, but I am still pursuing Benjamin Button.
As an aside: Whatever I am doing, and there are many variables that are not constant, I have had the same results from the Levine-based spreadsheet for almost a year. If I havenāt stopped aging, I am certainly slowing it down.
My problem with aging.ai is that it is not transparent. I understand the methodology for Levine. It has its issues with each biomarker, but it is coherent and rational.
The underlying difficulty is that if you are trying to create an algorithm to measure age related health (aging) then correlating it against chronological age is unhelpful.
As people know I think the key drivers are senescent cell load and ATP/O efficiency of mitochondria. That is actually a multidimensional issue which can be approximated to a single figure.
Aging.ai didnāt like giving me a 33 year discount. The updated algorithm says 14 year discount now using the same data as before. Thatās now within 1 year of the Levine model, which is oddly comforting. I admit it does feel weird being the same age as old people. I am the age of my grandmother when I first remember meeting her.
I wouldnāt read much into this. While these calculators are interesting and somewhat useful to get a rough idea of where you stand relative to your age, they are mostly useless for tracking changes in health or determining whether some change you did was good or bad.
I tend to agree with you on this. People need to know what their biomarkers mean individually and what changes in them means about their health. It is interesting and a little bit useful to put them into tests such as the Levine test. I am less sure about aging.ai because it is so opaque as an algorithm.
" I am less sure about aging.ai because it is so opaque as an algorithm."
True, but on the other hand, they do have a huge database that is constantly growing, and if they are not lying, learning from the input of users. I have been taking both of the tests, the Levine spreadsheet, and the Aging.ai calculator. In the beginning, the Aging.ai calculation was quite different from the Levine calculation, but over time it has come into close agreement with the Levine calculation, at least in my case.
Aging.ai no longer available, and an update from Alex: @biogerontology
āAlex Zhavoronkov, PhDā via Gerontology Research Group Sun, Jan 14, 11:01 AM (22 hours ago)
the data is depersonalized and was not used. And Aging.ai algorithms were simply old. I first published them in 2016.
Deep Longevity developed much better algorithms that have been deployed in many clinics and does not want to support the old platforms.
In cases where the system is credibly used for research purposes with the intent to publish, it can be used via API at scale and often free. Just contact their CEO, Deepankar Nayak
I hope to release a much more sophisticated Precious3GPT multi-modal transformer system that can work with multiple data types from different species including monkeys soon.
