These are directions I keep returning to. Some become projects. Most are research prompts.
human digital twin as a "prevention simulator"
A true digital twin should answer:
- What is changing in my physiology right now?
- Why might it be changing?
- What happens if I change sleep, nutrition, stress, activity, meds, or environment?
- What intervention is most likely to help, and how soon would I expect to see a signal?
real-time health visualization that drives action
Most dashboards show data. I want dashboards that drive decisions:
- "You're trending toward risk" → "Here's the smallest next step" → "Here's what to monitor."
LIFE wearable → clinically meaningful signals
Wearables are often noisy. The goal is:
- Better signal quality + artifact handling
- Physiologic inference (not just raw metrics)
- Explainable summaries clinicians can trust
- Continuous monitoring that respects privacy
N-of-1 trials for everyday life
Personalized medicine can start with:
- "Single-person trials" (structured self-experiments)
- Clear endpoints
- Lightweight protocols
- A model that updates as evidence accumulates
"prevention operating system"
A prevention OS would combine:
- longitudinal data
- clinical standards (FHIR/OMOP)
- coaching workflows
- simulation/forecasting
- and a feedback loop that adapts to real life
digital twin for longevity R&D
Connect the pipeline:
biomarkers → mechanisms → targets → interventions → outcomes
…and make it practical for:
- early-stage hypothesis testing
- trial design support
- phenotyping + stratification
- real-world evidence loops
disease prevention as a product, not a lecture
Most people don't need more information. They need:
- frictionless workflows
- behavior design
- accountability loops
- and outcomes that feel tangible
privacy-preserving health intelligence
I'm interested in architectures that keep trust intact:
- minimum data exposure
- local/on-device inference when possible
- strong audit trails
- clear user control over sharing
health coaching amplified by AI (but still human)
AI should reduce admin work and improve follow-through:
- summaries, plans, check-ins, personalization
- not replace the human relationship
big questions I'm obsessed with
- What does "health" mean quantitatively?
- Which signals change earliest before disease?
- How do we validate wearables clinically without slowing innovation?
- How do we make personalized prevention affordable at scale?
- What does the future of medicine look like when simulation becomes normal?