Hospital-grade signals are moving home.
Over-the-counter glucose biosensors, watches, phones, and open fNIRS ecosystems make dense physiology cheaper to collect.12
The bottleneck
Aging science is flying blind.
The field can often ask bold questions, but it cannot cheaply answer the most important one: did the intervention change the speed of aging?
-
1
Waiting for disease or death is too slow.
Hard clinical endpoints can take years or decades, so many promising ideas never get tested properly.
-
2
Most biological age tests are one-shot lab snapshots.
They can be expensive, sparse, and not yet trusted as regulatory-grade endpoints.
-
3
Small trials cannot rank thousands of interventions.
The result is guesswork: scattered experiments, weak comparability, and slow adoption.
PACE thesis
Measure the pace of aging cheaply, continuously, privately, and at population scale. Then every intervention can be judged in months instead of decades.
A
Try privately at home.
B
Train locally or federated across devices.
C
Report only proofs, model updates, and validated summary signals.
D
Pay contributors and inform the field.
Why now
Four things became true at the same time.
PACE is not betting on one magic biomarker. It combines cheap sensing, modern AI, privacy-preserving training, and a clearer validation rulebook.
Models can learn from messy streams.
New aging biomarkers already use high-dimensional signals such as methylation, retinal images, and dense digital data rather than a single manual feature.34
Learning can happen without pooling raw data.
Federated learning has become a practical architecture for training health models across sites while limiting raw-data movement.5
The field has validation criteria.
The Biomarkers of Aging Consortium proposed criteria for judging whether an aging biomarker is useful for longevity interventions.6
100xTarget reduction in time and cost per intervention readout.
LabsRank treatments quickly instead of guessing from weak proxy signals.
RegulatorsGet a yardstick that can be validated against outcomes and rulebooks.
PeopleEarn for private participation and get earlier warning once validated.
The program
$30-50M over 5-6 years.
A normal ARPA-scale infrastructure program: sensors at scale, contributor payments, and focused validation studies that decide whether to proceed.
1
Year 1
Signal
Recruit 1,000 people and ask whether cheap at-home sensors beat calendar age and catch a short-term stress-then-recovery swing.
- Go
- Beat baseline and detect the known short-term signal.
- No-go
- Sensors cannot separate signal from noise.
- Anchor
- Validate against best available lab tests in subgroups.
2
Years 2-3
Responsiveness
Scale to 50,000 people on the private network and test whether PACE detects a proven intervention, such as calorie restriction, in months.
- Go
- Detect the effect and retain contributors.
- No-go
- The model cannot track intervention response.
- Anchor
- CALERIE-style randomized evidence and biological-age assays.7
3
Years 4-6
Infrastructure
Open the validated test and dataset so any team can rank candidate treatments by anti-aging effect.
- Win
- The field adopts PACE as a standard readout.
- Proof
- Rank multiple interventions correctly against external evidence.
The product is not a consumer diagnosis at launch. Until clinically validated, PACE should be a research endpoint and trial infrastructure, not an unproven personal score.
Failure modes
Every honest moonshot needs gates.
The proposal is strongest when it treats failure as measurable and builds safeguards into the design.
Sensors are too noisy.
Combine many repeated signals instead of trusting one reading, then benchmark against lab assays in subgroups.
The model learns birthday or lifestyle.
Train against future health outcomes, known intervention responses, and validation criteria rather than chronological age alone.
People do not share enough data.
Make contribution private and paid: try privately, prove usefulness, report, and earn.
Biometric data or scores are abused.
Keep raw data on-device, delay public personal scores until validation, and govern access so it cannot be used to deny care, insurance, or employment.
Founder fit
Why you.
The proposal connects hardware grit, AI validation experience, and a direct obsession with making hard biological claims measurable.
"I built a working near-infrared spectrometer from open schematics and collected what may be the first open CGM plus fNIRS dataset."
- Think2earn hardware: cold-DMed funders, found a Vanderbilt Bowden Lab collaboration, and moved from schematics to real data.
- Model validation: fine-tuned vision transformers at a Forensics Institute to match trained experts on fake-video detection.
- Cross-domain background: two master's degrees, in Forensic Science and Information Studies, with practice turning messy evidence into hard ground truth.
- Next direction: build the infrastructure that lets the longevity field test faster, cheaper, and more honestly.
Clickable sources
Claims to anchor the proposal.
These are the source links behind the major factual claims in the page.
Stelo by Dexcom
Consumer-facing over-the-counter glucose biosensor evidence for the home-sensor premise.
Open sourceOpenfNIRS
Open functional near-infrared spectroscopy tools and hardware/software ecosystem.
Open sourceDunedinPACE
DNA-methylation biomarker designed to quantify the pace of biological aging.
Open sourceRetinal age gap
Deep-learning retinal-age gap associated with mortality risk in UK Biobank images.
Open sourceFederated learning in digital health
Review of federated learning as a privacy-preserving architecture for health AI.
Open sourceBiomarkers of aging criteria
Biomarkers of Aging Consortium work on identifying and evaluating longevity-intervention biomarkers.
Open sourceStress and recovery
Study reporting that biological age can rise with stress and fall after recovery.
Open sourceCALERIE calorie restriction
Randomized-trial evidence on long-term caloric restriction and DNA methylation measures of biological aging.
Open sourceFramingham Heart Study
NHLBI overview of the long-running cohort that helped define modern cardiovascular risk factors.
Open sourceUK Biobank
Open-access cohort/resource paper describing UK Biobank as a large-scale health research resource.
Open sourceValidation of aging biomarkers
Nature Medicine perspective on validating biomarkers for longevity-intervention use.
Open source