Biological Age

....

 

There is a tradition of algorithmic tools to predict biological age and lot has been published. One of the best I found is the BAS (Biological Age Score of Klemera and Doubal) as described e.g. by Levine which also includes functional and other tests as pulmonary FEV, systolic blood pressure and CMV-cytomegalovirus presence). Klemera and Doubal method (KDM) was shown to reliably predicting mortality on a large cohort.

 

Levine ME. Modeling the rate of senescence: can estimated biological age predict mortality more accurately than chronological age?. J Gerontol A Biol Sci Med Sci. 2013;68(6):667-74.

My apologizes, I just realize the link of Morgan Levine's paper I quote does not work, here is the correct one:

https://www.ncbi.nlm...pubmed/23213031
 

And as I am on it, let me quote or re-quote here a more recent paper by Morgan Levine et al you might have seen already:

An epigenetic biomarker of aging for lifespan and healthspan

https://www.ncbi.nlm...pubmed/29676998

(edit: second paper)

Edited by albedo, 03 September 2018 - 09:36 PM.

Any additional input on this very important topic? Is there another thread which I can follow more fruitfully?

As a teaser, here is an excellent data base where, to exemplify a possible use, I selected tissue (here "blood") changes with age: http://ageing-map.or.../tissue/200012/

"The Digital Ageing Atlas (DAA) is a portal of age-related changes covering different biological levels. It integrates molecular, physiological, psychological and pathological age-related data to create an interactive portal that serves as the first centralised collection of human ageing changes and pathologies."

Jumping in on the question of epigenetic aging clocks.  There are actually several purported epigenetic aging clocks out there. The one that everyone talks about, for some reason, is the first-generation Horvath clock; it's available commercially. It is tightly correlated with chronological age, not biological age, so it isn't a good way to evaluate your biological age or the effects of a potential anti-aging intervention.
 
The new Horvath epigenetic aging clock, DNAm PhenoAge (PMID 29676998) does closely track biological age, having been built up by machine learning from a series of (mostly) common blood test analytes; it would be a good test to use to evaluate putative anti-aging interventions; unfortunately, it's not commercially available. (Additionally, as the paper notes "In fact, but perhaps not surprisingly, the phenotypic age measure used to select CpGs [ie, the composite of underlying blood and other tests] is a better predictor of morbidity and mortality outcomes than DNAm PhenoAge." Unfortunately no one has turned this into a handy online calculator or app, though I wish someone would ...).

You can see Horvath's presentation and Undoing Aging on this.

@Michael.

I am glad you have found a little time to follow this subject also on this Forum as I have already expressed here which to me also tells the importance of the subject. And I am also happy as this seems the right place to discuss the topic.

The work by Levine et al. is one of the most interesting I found so far for the very reasons you mention and you also answer a question I have since time about the commercial availability of the DNAm PhenoAge which hopefully might change in the future.

I feel more progress will also happen when a system biology approach will be taken to the biological age determination, possibly integrating several approaches such as the epigenetic, clinically biomarkers, anthropometric and machine learning based. IMHO this is much expected as, similarly, a system biology and multi-level approach to aging is important for a better understanding. Morgan Levine replies a bit in this sense to a critic by Mitnitski and Rockwood to a previous paper of her:

"... I agree with Mitnitski and Rockwood that a systems biology approach is important, and in moving forward, algorithms need to incorporate interactions between various systems/levels, which may rely on more advanced computational techniques—such as machine learning...."

Levine ME. Response to Dr. Mitnitski's and Dr. Rockwood's letter to the editor: Biological age revisited. J Gerontol A Biol Sci Med Sci. 2014;69(3):297-8.

Hopefully you will keep feeding this thread :-)

Jumping in on the question of epigenetic aging clocks.  There are actually several purported epigenetic aging clocks out there. The one that everyone talks about, for some reason, is the first-generation Horvath clock; it's available commercially. It is tightly correlated with chronological age, not biological age, so it isn't a good way to evaluate your biological age or the effects of a potential anti-aging intervention.
 
The new Horvath epigenetic aging clock, DNAm PhenoAge (PMID 29676998) does closely track biological age, having been built up by machine learning from a series of (mostly) common blood test analytes; it would be a good test to use to evaluate putative anti-aging interventions; unfortunately, it's not commercially available. (Additionally, as the paper notes "In fact, but perhaps not surprisingly, the phenotypic age measure used to select CpGs [ie, the composite of underlying blood and other tests] is a better predictor of morbidity and mortality outcomes than DNAm PhenoAge." Unfortunately no one has turned this into a handy online calculator or app, though I wish someone would ...).

 

You can see Horvath's presentation and Undoing Aging on this.

Great so that was a waste of time and mostly worry getting my Horvath "myDNAge" test done and having results that were 6 months older than my chronological age.

I thought Horvath was against undoing aging because of some tired old rebutted arguments.

This is quite old as out of a workshop of expert in UK in 2012 I wonder if someone of you know something along the same lines (more systemic) but recent and with longitudinal tracking:

Lara J, Cooper R, Nissan J, et al. A proposed panel of biomarkers of healthy ageing. BMC Med. 2015;13:222.

 2012 BM Newcastle.PNG   446.01KB   0 downloads

...
The new Horvath epigenetic aging clock, DNAm PhenoAge (PMID 29676998) does closely track biological age, having been built up by machine learning from a series of (mostly) common blood test analytes; it would be a good test to use to evaluate putative anti-aging interventions; unfortunately, it's not commercially available....

Yes. Also, from my reading, it is quite easy to measure your "phenotypic age" (Step 1 in the paper) based on your own biomarkers and the Gompertz proportional hazard model coefficients but I wonder which practical use you can make of the values gives in Table S6 which build up the model (Step 2 in the paper) using the DNA methylation sites to predict the "phenotypic age" (what they call "DNAm PhenoAge).
 

Great so that was a waste of time and mostly worry getting my Horvath "myDNAge" test done and having results that were 6 months older than my chronological age.

 

I thought Horvath was against undoing aging because of some tired old rebutted arguments.

 Not sure about that. Aren't we really talking about methylation age vs chronological age? So you could be 40 yrs old, but you have methylation of a 45 yr old (not so good). I think that's good to know. Methylation has to have some correlation to biological age, or at least it would seem so. Still, confusing to say the least.

Surely this is the holy grail -- to find one test that can accurately assess your biological age.  Would seem that before that can happen, scientists need to find some holistic theory that comprehensively explains aging.  Not there yet.

Am intrigued by the Aging.AI approach. Sure would be easier if they recommended one test for all the required biometrics.

Also intrigued by Dr. Peter Attia's 5 blood tests he suggests everyone takes, which ties into his piece about how to increase lifespan.

Yes. Also, from my reading, it is quite easy to measure your "phenotypic age" (Step 1 in the paper) based on your own biomarkers and the Gompertz proportional hazard model coefficients but I wonder which practical use you can make of the values gives in Table S6 which build up the model (Step 2 in the paper) using the DNA methylation sites to predict the "phenotypic age" (what they call "DNAm PhenoAge).
 

Hi albedo...

I understand that your post contains important content. I see and appreciate your effort to support and trigger progress.

I believe we're on parallel paths. You're ahead of me in some ways, I'm ahead of you in other ways...

So, of course, I'm interested most in learning more about the ways you're ahead.

Your post has intriguing content in it...

• I don't understand its full meaning... Specifically, I don't understand what you mean to say beginning with the phrase "... but I wonder". Could you expand on what you mean to say?
  • '... which practical use..." means what?
• I take your post to suggest that a few or many or most of the calculations required to calculate DNAm PhenoAge might be easily calculated. Am I understanding you correctly? If yes, do you have the knowledge and experience needed to create an Excel spreadsheet Draft of the data needed to do the calculations?
  • If you, or anyone, can correctly express the objects in a draft version 0.01 of an Excel spreadsheet, I can move to take it and get it to draft version 0.02 and we're off to the races.

You've seen Vincent Giuliano's most recent post right? He appears to suggest DIY Self-Hacked ReProgramming is in the Evolutionary Cards...

Best!

Steve

Hi albedo...

 

I understand that your post contains important content. I see and appreciate your effort to support and trigger progress.

 

....

Thank you Steve! I enjoy to help!

 

I don't understand its full meaning... Specifically, I don't understand what you mean to say beginning with the phrase "... but I wonder". Could you expand on what you mean to say?

• '... which practical use..." means what?

   

Yes, I was not particularly clear. Sorry. What is calculated in Step 1 of the Levine's paper is what they call "phenotypic age". In Step 2 they regress the cohort DNA methylation data on the phenotypic age. I can with some math calculate my phenotypic age and the risk of mortality from my biomarkers (see after). With "practical use" I only meant whether of not I can extract from the Step 2 and the supplementary material provided (in particular the Table S6) some information on my own DNA methylation data w/o taking a test. Very likely not and I admit I need to understand better this point. You might also look at this presentation:

http://gero.usc.edu/...AA/1.Levine.pdf

 Levine 2018 DNAm PhenoAge.PNG   93.66KB   0 downloads

 

I take your post to suggest that a few or many or most of the calculations required to calculate DNAm PhenoAge might be easily calculated. Am I understanding you correctly? If yes, do you have the knowledge and experience needed to create an Excel spreadsheet Draft of the data needed to do the calculations?

Yes indeed I do have a spreadsheet calculating both the mortality risk and the Phenotypic Age based on the Step 1 biomarkers. I developed this in parallel to another great poster in another Forum you might refer to. However his version is much more elegant and clean than mine and so I refer you directly to his post HERE. I could cross check all his calculation with my calculation to avoid mistakes and we are discussing also other interesting implications.

 

You've seen Vincent Giuliano's most recent post right? He appears to suggest DIY Self-Hacked ReProgramming is in the Evolutionary Cards...

I did have seen the one in the In-vivo reprogramming thread but I did not research more on him. Please let me know maybe in the other thread.

In an interesting thread related to Heat Shock Protein, iPSC and epigenetic reprogramming, HighDesertWizard provided some of the results of several LongeCity Members participating to the Longecity biological age program using different methodologies:

 

....

 

Longecity has an aging biomarkers program and about 20 people are participating in it. Here's a spreadsheet graph of their results.

 

 

Notice...

• there are several Longecity members with a lower MyDNAge test result than their actual age
• let's assume the MyDNAge test result provides rough test result comparability to the Grant Age Estimation algorithm.
• there are 6 members highlighted in yellow with an estimated age score significantly lower than their actual age
• this is anecdotal data of course, that's a given
• but if HSP expression is really as key to epigenetic age, shouldn't member descriptions of their practices include references to HSP promoting activities?

https://www.longecit...ndpost&p=861944

Likely, many of those participants have large experience with interventions and equally likely keep a record of the relevant clinical biomarkers upon which is based the new Levine’s and Horvath’s clock recently published (An epigenetic biomarker of aging for lifespan and healthspan ) potentially tracking biological age. It would be interesting if they could compare with the Phenotypic Age as calculated by JGC and independently confirmed by myself here :

https://forum.rescue...-phenotypic-age

https://www.dropbox....ge_gen.xls?dl=0

Following the Phenotype Age determination (and Mortality Risk), Levine et al. also regressed the Phenotype Age on blood DNA methylation data of a particularly well characterized cohort (InCHIANTI), determining a new precise epigenetic biomarkers of aging, aka DNAm PhenoAge. The spreadsheet I reported in the above links allows a calculation of the Phenotype Age and just a rough estimation of the DNAm PhenoAge by fitting the regression data directly from the plot (see the JGC’s post for details) given in the Levine’s presentation:

 Levine 2018 DNAm PhenoAge.PNG   93.66KB   1 downloads

http://gero.usc.edu/...AA/1.Levine.pdf

Anecdotally my data are:

Age: 63 years

Phenotypic Age: 46.1 years

Mortality Risk (10 years): 0.033

DNAm PhenoAge (rough estimate): 45.6 years

Something which is still unclear is the relative biological role the several clinically relevant biomarkers play in the determination of the Phenotype Age as discussed in the RescueElders Forum thread, e.g. the high weight of RDW and MCV when compared to inflammatory markers such as hr-CRP. You might wish to read the discussion also on that Forum. I am trying to better understand this point.

Interestingly, my data are extremely similar to what obtained using Aging.ai (V1.0 only) and also a ML/AI trained system on face images. Where the methodologies seem diverging is in the time trends of “biological age” so calculated which I am still understanding, the trends being more important than the absolute values when we want to track interventions, IMHO. But this will be for another post….

(edit: spelling)

Edited by albedo, 12 November 2018 - 04:51 PM.

"Biomarkers of the physiological state and biological age of individuals.

Biomarkers in human research are, on the one hand, used to detect individual variability in the progress of ageing, as risk indicators, and, on the other hand, for monitoring the response to
interventions. Different biomarkers have been developed to answer different questions, for example, to monitor the physiological state of individuals, predict the onset and/or progression of age-related diseases, detect the physiological vulnerability of elderly to poor clinical outcomes or predict mortality. Biomarkers of the risk of age-related diseases have been developed with great success. No consensus has yet been reached on biomarkers of biological age, that is, the mismatch between chronological age and the stage of an individual along the ageing process. These biomarkers should ideally meet a number of criteria, such as those defined by the American Federation for Aging Research (AFAR): they should (1) mark the individual stage of ageing and predict mortality better than chronological age; (2) monitor ageing in a range of systems and not the effects of disease; and (3) allow longitudinal tracking (for example, by blood tests or imaging techniques) in animals and humans171. Several types of biomarker of the physiological state include whole-system indicators of physical or mental capability (for example, locomotor function, strength, balance, cognition and activity during daily living), physiological reserve (for example, respiratory and cardiovascular function) and the systemic capacities to regulate lipid and glucose metabolism and immunity (for example, insulin, IL-6 and CRP)33,172. In addition to single markers, multi-marker indicators have been generated based on assays of multi-organ functionality and/or molecular characteristics. Physiological vulnerability later in life, that is, ‘frailty’ at ages above 80 years is generally described by low physical activity, muscle weakness, slowed performance, fatigue or poor endurance and unintentional weight loss. About 50 different frailty algorithms are available, the ‘frailty phenotype’173 and ‘frailty index’174 being the most commonly used clinically. For early phases of life, other scores, such as the ‘Pace of Aging’ score43, have been generated. More recently, multi-marker indicators of biological age have been based on age-related changes in the transcriptome175, epigenome176,177, metabolome178 and structural neuroimaging179. These await systematic testing and comparison with each other and with traditional parameters, in relation to clinical decisions and intervention studies. Different indicators of biological age (telomere shortening, epigenetic clocks and pace of ageing) seem to reflect different aspects of physiological decline180. Because long-lasting cohort studies contain many ageing phenotypes and a large amount of clinical, imaging and molecular data, collected at multiple time points, these studies could allow systematic comparisons and development of a multivariate mix of marker profiles with the strongest predictive power."

Did someone studied LP(a) as a possibly proxy of IL-6 I never measured? Thanks.

Partridge L, Deelen J, Slagboom PE. Facing up to the global challenges of ageing. Nature. 2018;561(7721):45-56.

Excellent and actionable expert panel review and recommendation of BMs of aging, also exemplified by the TAME trial on metformin:

Justice JN, Ferrucci L, Newman AB, et al. A framework for selection of blood-based biomarkers for geroscience-guided clinical trials: report from the TAME Biomarkers Workgroup. Geroscience. 2018

 BM Aging.PNG   214.98KB   1 downloads

Several BMs are known to many of us. Note also at the bottom of the picture the emergence of epigenetic, multi-omic BMs possibly capturing complex underlying overall aging processes.

Edited by albedo, 08 December 2018 - 04:34 PM.

...

Something which is still unclear is the relative biological role the several clinically relevant biomarkers play in the determination of the Phenotype Age as discussed in the RescueElders Forum thread, e.g. the high weight of RDW and MCV when compared to inflammatory markers such as hr-CRP. You might wish to read the discussion also on that Forum. I am trying to better understand this point.

...

Something discovered recently, based on a reply received directly by Dr. Levine to my questions, is that while recognizing biomarkers such as RDW (geometric) have a surprisingly high impact on the estimate of phenotypic age, the weights in the paper's Table 1 which enter the phenotypic age calculation (step 1) and then the regression with the DNA methylation data (step 2) are NOT standardized to allow for a direct comparison. This is something I would expect to be considered on future follow on works.
 

Something discovered recently, based on a reply received directly by Dr. Levine to my questions, is that while recognizing biomarkers such as RDW (geometric) have a surprisingly high impact on the estimate of phenotypic age, the weights in the paper's Table 1 which enter the phenotypic age calculation (step 1) and then the regression with the DNA methylation data (step 2) are NOT standardized to allow for a direct comparison. This is something I would expect to be considered on future follow on works.
 

Can you clarify what you mean in the above? What exactly is not standardized — the RDW measurement? Surely that is pretty standardized. Or something about the regression step? How can that be?

Can you clarify what you mean in the above? What exactly is not standardized — the RDW measurement? Surely that is pretty standardized. Or something about the regression step? How can that be?

It might mean this: "In statistics, standardized [regression] coefficients, also called beta coefficients or beta weights, are the estimates resulting from a regression analysis that have been standardized so that the variances of dependent and independent variables are 1.^[1])
 

https://en.wikipedia...zed_coefficient

Does anyone here have a clue why the v 1.0 and v 3.0 of Insilico Medicine's Aging.Ai predictor differ so largely? In my case v 3.0 is systematically lower than v 1.0, in average -33%

I only did a superficial reading of the two different references given for v 1.0 and v 3.0 so maybe I am missing a crucial point. Thank you !

Does anyone here have a clue why the v 1.0 and v 3.0 of Insilico Medicine's Aging.Ai predictor differ so largely? In my case v 3.0 is systematically lower than v 1.0, in average -33%

 

I only did a superficial reading of the two different references given for v 1.0 and v 3.0 so maybe I am missing a crucial point. Thank you !

Why did they not include homocysteine?

Why did they not include homocysteine?

Good point. Not sure. Maybe because not really common to be prescribed in a lab work? In any case I am skeptic this can explain the difference. I tend more to think about the machine learning steps in use or?
 

Thinking about senolytic activity and biomarkers of aging I pop into this (a bid old though):

"... At the level of cells, a potential biomarker of aging may be the presence of  senescence. Senescence is a condition in which old or damaged cells remain alive but cease to reproduce. This is an important tool in the body’s ability to prevent cancerous tumors from developing. The older a person becomes, the more senescent cells he or she accumulates. Several markers of senescence in humans have been suggested as biomarkers of aging. On November 10, 2011, researchers at the Mayo Clinic published a study in the journal Nature showing that an accumulation of senescent cells may lead to age-related diseases, at least in animals. By removing most of these cells from several organs (body fat, eye, and skeletal tissue) of lab mice, the investigators were able to significantly delay the onset these diseases, or stop their progression if they had already become established. The study suggests what may prove fruitful areas for future researchers to explore in the search for true biomarkers of aging..."

https://www.afar.org...AGING_2016.pdf

Comments?

Thinking about senolytic activity and biomarkers of aging I pop into this (a bid old though) ...

And the cited paper using p16 as potential biomarker:

Baker DJ, Wijshake T, Tchkonia T, et al. Clearance of p16Ink4a-positive senescent cells delays ageing-associated disorders. Nature. 2011;479(7372):232-6.
https://www.nature.c...les/nature10600

Thinking about senolytic activity and biomarkers of aging I pop into this (a bid old though) ...

Going back to Aubrey's book Ending Aging (Chapter 10, pp 236-237) he gave a possibly clue (bold mine) looking at senescence-associated beta galactosidase (SA-beta-gal) plus p53, 53BP1, p21 and p16 levels:

"...What's emerging, then, is a picture of SA-beta-gal as an enzyme that appears at high level in the main bodies of cells undergoing some kind of stress that may ultimately threaten their neighbors. This might mean that by using high levels of SA-beta-gal as an identifier for the destruction of senescent cells, we would simultaneously take out some useful "targets of opportunity." However, we may be able to establish a system of double-checks, to help us weed out more genuinely senescent cells while leaving more innocent (but suspicious-looking) cells unmolested. This is because, in addition to SA-beta-gal, senescent cells also produce abnormally high levels of other molecules involved in the programmed senescence response. Senescent baboon skin cells, for example, contain an activated form of the protein ATM kinase, which responds to DNA damage by activating several tumor suppressor genes, including the famous p53. Senescent cells also exhibit high levels of p53, as well as the binding protein (53BP1) by which its gene interacts with ATM kinase, and p21, a senescence regulator that works under 56 p53's command. Some senescent cells also contain high levels of p l 6 , the other main regulator of the process. Levels of this protein, for reasons as yet unknown, also climb slowly with age in nonsenescent cells, making it an unreliable marker for senescence when taken in isolation; but it—like these other features—could still potentially be used as part of a double-checking mechanism, with multiple proteins being used to distinguish genuinely senescent cells from those expressing only one of them for unrelated reasons...."
 

Adding on how to test for effectiveness of senolytics self experimentation, these two posts by Reason provide some detail and include also biological age determination:

https://www.fightagi...ts-and-measures

https://www.fightagi...rug-candidates/

"...The number of possible options grows on a month by month basis, but it may be that, at this stage, more effort should go towards calibrating the behavior of an existing biomarker approach, following use of interventions to slow or reverse aspects of aging, rather than continuing to pile additional markers onto the heap..."

Selection of Recent Research into Biomarkers of Aging

https://www.fightagi...rkers-of-aging/

I was just waiting for this to happen. Here is in my view a quite pioneering work on biological age determination using ML/AI on microbiota. The study is still at level or preprint as per today:

"...Our most accurate DNN regressor achieved the MAE of 3.94 years. This performance is comparable with the 1.9 MAE of the PhotoAgeClock, 2.7 of the state of art methylation aging clock, 7.8 MAE transcriptomic aging clock and 5.5 MAE of the hematological aging clock published previously. We also developed a method for microbiological feature selection and annotation..."

Quite fascinating is that: "...Interestingly, while it contains both beneficial (e.g. Bifidobacterium) and pathogenic (e.g. Pseudomonas aeruginosa) microbes, seno-positive or seno-negative status is not determined by the nature of host-microbe interactions (Figure 12)..."

bioRxiv 507780; doi: https://doi.org/10.1101/507780

The (great) Levine et al paper, specifically on Phenotypic Age (and Phenotypic Age Acceleration) based on blood biomarkers (and age) is now out (Dec 31, 2018) validated on the NHANES IV cohort. See:

Liu Z, Kuo PL, Horvath S, Crimmins E, Ferrucci L, Levine M. A new aging measure captures morbidity and mortality risk across diverse subpopulations from NHANES IV: A cohort study. PLoS Med. 2018;15(12):e1002718.

https://journals.plo...al.pmed.1002718

"...Phenotypic Age was calculated based on a linear combination of chronological age and 9 multi-system clinical chemistry biomarkers in accordance with our previously established method. We also estimated Phenotypic Age Acceleration (PhenoAgeAccel), which represents Phenotypic Age after accounting for chronological age (i.e., whether a person appears older [positive value] or younger [negative value] than expected, physiologically). All analyses were conducted using NHANES IV (1999–2010, an independent sample from that originally used to develop the measure)..."

This follows on the previous paper "An epigenetic biomarker of aging for lifespan and healthspan" which is also discussed in this thread as potentially actionable to us, see: https://www.longecit...ndpost&p=862525

In case you overlooked this, it is 7.5 k$ though! I expect some clinics will buy it and re-sell as a service. I recollect in a couple of occasion having used the H-Scan predecessor.

https://www.agemeter..._eid=44a6ef5d7b

The (great) Levine et al paper, specifically on Phenotypic Age (and Phenotypic Age Acceleration) based on blood biomarkers (and age) is now out (Dec 31, 2018) validated on the NHANES IV cohort. See:

 

Liu Z, Kuo PL, Horvath S, Crimmins E, Ferrucci L, Levine M. A new aging measure captures morbidity and mortality risk across diverse subpopulations from NHANES IV: A cohort study. PLoS Med. 2018;15(12):e1002718.

https://journals.plo...al.pmed.1002718

 

"...Phenotypic Age was calculated based on a linear combination of chronological age and 9 multi-system clinical chemistry biomarkers in accordance with our previously established method. We also estimated Phenotypic Age Acceleration (PhenoAgeAccel), which represents Phenotypic Age after accounting for chronological age (i.e., whether a person appears older [positive value] or younger [negative value] than expected, physiologically). All analyses were conducted using NHANES IV (1999–2010, an independent sample from that originally used to develop the measure)..."

 

This follows on the previous paper "An epigenetic biomarker of aging for lifespan and healthspan" which is also discussed in this thread as potentially actionable to us, see: https://www.longecit...ndpost&p=862525

albedo... Thanks for posting this study by Levine / Horvath.

• I've been on the lookout for this kind of study for a while. That is, I've been looking for solid studies that focus on Survival Probability in the future given some state of health.

I think it's important to highlight a few of the graphic figures of the study.

This last pic is especially interesting. The Area Under the Curve (AUC) is a measure of the sensitivity/impact of the independent variable on Survival Probability.

It makes clear that fighting specific diseases we find ourselves having is an important objective.

...

It makes clear that fighting specific diseases we find ourselves having is an important objective.

Yes, and we might add the Table 1 which makes your point more explicit: look e.g. at the importance to have sugar control under check (diabetes) and well functioning kidneys (nephritis/nephrosis) both with a HR of ~20%. We knew this already but what is important here is Phenotypic Age is very well predictive of this, because of its construct, and can possibly guide specific tests and successive treatments in the clinical setting: as the authors write (bold mine):

"...we found that Phenotypic Age was predictive of disease-specific mortality including heart disease, cancer, chronic lower respiratory disease, diabetes, influenza/pneumonia, and nephritis/nephrosis, with exception of cerebrovascular disease mortality (HR = 1.03, 95% CI = 0.98–1.09). HRs were the highest for diabetes and nephritis/nephrosis, suggesting that a 1-year increase in Phenotypic Age relative to chronological age increases the risks of death from these causes by about 20%. For the other causes (aside from cerebrovascular disease), a 1-year increase in Phenotypic Age increased risk by between 7% (cancer and chronic lower respiratory disease) and 12% (influenza/pneumonia)..."
 

 disease - phenotypic.PNG   299.22KB   1 downloads