I have been deeply involved in biohacking, quantified self, and consumer health tech for over 10 years. Until recently, I handled most of my data in Excel, but recent developments in AI have led me into the world of vibe coding. I’ll describe my project here, and I’d be very interested to hear feedback, comments, questions, and sparring from others, especially from people working on similar projects.
I also work closely with an elite athlete, so this project is relevant for him as well.
General overview of the project
Goal:
For me, the goal is to build an in-depth understanding of health, wellbeing, and longevity, personalized to individual goals and needs, and grounded in the latest science.
For my colleague, the goal is to build an in-depth understanding of elite-athlete-level performance and recovery, personalized to individual goals and needs, and connected to the latest science. Done right, this could become one of the deciding factors in winning Olympic gold.
For now, this is only for personal use.
Current tools
Codex + ChatGPT 5.5 + Hermes agent
Phase 1: Data integration and dashboard
The first phase is about combining all the data in the backend and visualizing the most relevant data in the frontend dashboard.
Integrations created so far:
-
Oura Ring: sleep, recovery, and health
-
Garmin: activity and sport-related data
-
Withings: body composition and health-related data
-
Airthings: bedroom environment
-
Subjective metrics logged via phone
Manual data uploads:
-
Subjective metrics, including historical data
-
Aktiia 24h blood pressure measurements, almost 20,000 measurements
-
Nuanic electrodermal activity measurements
-
Blood tests
-
Phone usage
-
Maternity and Child Health Clinic information
-
Medical records
-
Plus much more
Currently, the system includes 13 different data sources, 116 tracked metrics, and 31k+ data points.
In the future, the goal is to expand this to 30+ data sources, 300+ tracked metrics, and 1M+ data points.
Phase 2: Data analysis tools
This phase focuses on:
-
Automated insights
-
Correlation explorer
-
Future forecasting
A first quick forecasting model is already ready, but it still needs improvement.
Phase 3: N=1 module
This module will include all historical N=1 experiments. This part has not been started yet.
Other activities and considerations
-
Avoiding the “garbage in, garbage out” problem
-
Creating a value ranking for each metric
-
Building an AI agent that scouts relevant studies based on personal goals and targets, and highlights them as possible activities, interventions, or N=1 experiments
-
Data security considerations, currently everything is just local on my computer
-
Support for verbal and written daily diary entries
-
A visual one-page overview
Questions I would like this project to help answer
-
Which metrics predict future performance, rather than just describe the past?
-
What actually improves performance and recovery?
-
What separates average or good days from exceptional days?
-
Are we reaching individual goals or not?
-
Which interventions are worth keeping, changing, or removing?
-
What are the early warning signs of overreaching, illness, injury, or burnout?
-
How do we compare against other elite athletes, the same age group, or our younger selves?
-
What are the most relevant findings from the latest studies that should be tested or incorporated into our protocol?