Daniel Belsky

Aging, Genes, Environment, Life Course Epidemiology

[Daniel Belsky @Danbelsky]

@Experian_US I am unable to log in. Your electronic system will not send verification texts to my phone. Your password reset system claims my phone number does not match my account (which is not possible). You do not seem to offer any live customer support. Please advise asap.

[Daniel Belsky @Danbelsky]

@AcadBehavMedRes Humbled and honored to receive! Wonderful sharing our work with this academy. Such innovative and important research about how we can (and sometimes can’t) change behavior to improve health. Learned a lot!!

[Daniel Belsky @Danbelsky]

@DrtobeM @NIH @ColumbiaAging @DrRiveraMindt Thrilled to have @DrtobeM joining us @ColumbiaMSPH @ColumbiaAging !!

[Daniel Belsky @Danbelsky]

@AcadBehavMedRes Such an honor to receive this award (and apologies for the delayed response — have been away from Twitter)

[Xu Gao @XuGao11]

Here is my latest research with @Danbelsky
https://t.co/T5ErqhapRL on [#aging and #mentalhealth ], published with @NatureComms. Aging could promote incident depression and/or anxiety. @PKU1898 @ColumbiaMSPH @ColumbiaAging

[Daniel Belsky @Danbelsky]

@DrMorganLevine @KarlPfleger @davidasinclair @TruDiagnostic Figure shows changes in treatment and control. What we see is that CR and AL show the same pattern of change, contrary to the H that CR would slow change. We had a figure showing the effect estimates (more like what you want) but Eds had us cut them. They are in text and tables.

[Daniel Belsky @Danbelsky]

@DrMorganLevine @KarlPfleger @davidasinclair @TruDiagnostic Finally, re the nature of the effect: we tested if CR slowed the rate of increase in Grim/ Pheno. (Model tested diff between CR and control in change from baseline) So test was sensitive to an intervention effect that slowed change in Grim/ Pheno.

[Daniel Belsky @Danbelsky]

@DrMorganLevine @KarlPfleger @davidasinclair @TruDiagnostic Re why not residualize clocks: answer is b/c there is no natural setting in which to estimate the regression. Longer discussion than I have the Twitter chops for. But will say that we did run models with age residuals estimated from a variety of settings and result is unchanged.

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson Thanks for you work to aid interp of our study! (And all the others that you cover)

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson But I agree that, to the extent positive change in AL is noise (vs, eg regression to the mean among people who met inclusion criteria b/c they were living healthier than their typical lifestyle at the time of enrollment) this could inflate tx effect.

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson Again, the right measure of effect here is the “difference in difference”, ie the diff between within-person change in CR vs within-person change in AL. The purpose of this design is to take baseline diffs between groups out of the equation.

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson Should make PACE more sensitive to changes, but effects of life history can still be cumulative on the rate observed at pre-tx baseline

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson Might be worth considering duration of the “treatment”. That 2y of CR on top of healthy baseline has less impact vs decades of healthy living… seems plausible to me

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson Diff between PACE and, eg PC GrimAge is that PACE calculation includes a normalization step designed to make values more comparable to the Dunedin Study reference.

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson Clock ages at baseline are all over the place https://t.co/PwX1nQbawH

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson Also, not sure where you are getting your est of the tx effect from. Comparison needed is delta in the CR group vs delta in the AL group. Effect is still small, but more on the order of 1/2 the diff btw CALERIE at baseline and the Dunedin cohort.

[Daniel Belsky @Danbelsky]

@KarlPfleger @mike_lustgarten @olafurpall80 @bryan_johnson One difference is that the CALERIE participants were healthy volunteers selected to be non/obese and free of chronic disease whereas Dunedin cohort is representative of the population of NZ South Island.

[Daniel Belsky @Danbelsky]

@KarlPfleger @TruDiagnostic https://t.co/MqDrNtDa23 “The resulting Pace of Aging was then scaled to a mean of 1, so that it could be interpreted with reference to an average rate of 1 year of biological aging per year of chronological aging.”

[Daniel Belsky @Danbelsky]

@KarlPfleger @TruDiagnostic Yes. More specifically, the reference population for this calculation is the Dunedin Study birth cohort.