In this twentieth episode, Polina Mamoshina introduces recently launched Deep Longevity, and its app (young.ai).
Read the transcript
Biomarkers of aging are introduced. She explains that they have taken a superior approach by using deep learning instead of machine learning. Aging clocks in general are covered. Finally, she shares her view that transcriptomic and proteonomic clocks are the likely future.
Topics we discussed in this episode
Personal background: Moscow State University, Oxford University, Insilico Medicine hackathon
Bringing Deep Longevity out of stealth, Young.ai companion app
Deep Longevity introduction including company aims
Description of Young.AI app
Biomarkers of aging as the accelerant of market for aging interventions
Introduction to aging clocks: Horvath, Hannum
Taking a novel and superior technological approach to aging clocks, using deep neural networks, instead of shallow machine learning
Limitations of shallow machine learning models
Ability of neural networks to capture highly non linear dependencies and what that matters for biological age determination
Investing in anticipated payoff from deep learning over the long-term, even if machine learning may be good enough in many cases now
Biological age prediction with Aging.ai
Two approaches to designing aging clocks
Machine learned PhenoAge biological age score
Introducing mortality, with the GrimAge score
Longevity clinics and life insurance as market
Biological age scoring as onboarding tool for life insurance markets
Training datasets
Common blood analytes used in PhenoAge vs Aging.ai
Optimized blood analyte levels for a given individual to get younger
Orthodox medicine uses blood analyte levels that are not specific to the individual and not optimized ranges; designed to detect only late-stage pathologies
Cheapness of regular blood analytes
Emerging market is likely to age score bodily subsystems rather than provide an overall singular biological age score
Goal is to find the fastest ticking clock in your body
Biological age test using a selfie
Providing a library of biological age scores, from free to expensive, so users can upgrade, find out more about themselves
Belief that proteomic and transcriptomic clocks will outperform epigenetic clocks in terms of being actionable with interventions
Epigenetics and aging
Acceleration of the aging rate may show up "late" in terms of being able to intervene, on the epigenome
Youthful blood plasma exchanges and age quantification
Transcriptomic, proteomic, and glycomic clocks
Anticipated rise of longevity clinics
Show links
Deep Longevity (Company Website)
Insilico Medicine (Company Website)
Human Longevity, Inc. (Company Website)
Regent Pacific Group (Company Website)
Young.AI (App from Deep Longevity)
Aging.AI (Biological Age Prediction)
'DNA Methylation Age of Human Tissues and Cell Types' (Paper)
'Assessment of Epigenetic Clocks as Biomarkers of Aging in Basic and Population Research' (Paper)
Steve Horvath (WikiPedia Entry)
'Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates' (Paper)
Gregory Hannum (LinkedIn)
Morgan Levine (LinkedIn)
'An epigenetic biomarker of aging for lifespan and healthspan' (Paper)
Elysium Health (Company Website)
'DNA Methylation GrimAge Strongly Predicts Lifespan and Healthspan' (Paper)
FOXO BioScience (Company Website)
NHANES III (CDC)
AgeoTypes (Stanford Article)
GlycanAge (Company Website)
GENOS (Company Website)
