Turns out my body has been keeping receipts.
After feeding 28 days of health data into an LLM, the patterns that emerged feel obvious in hindsight — and that’s exactly the problem with dashboards. The data was always there. I just couldn’t see it through the charts.
The headline: sleep quality predicts next-day stress with uncomfortable accuracy. When restorative sleep (deep + REM as a percentage of total) drops below 30%, average stress jumps to 38. Above 45%? It’s 24. That’s not a minor correlation — it’s a 37% reduction.
The stress triggers are blunt:
- Low restorative sleep (<1h) → +12 stress points next day
- Missed sleep data → +15 stress points next day
- Short total sleep (<6h) → +8 stress points next day
Missed sleep data being the worst offender is almost poetic. The nights I forget to charge my watch or fall asleep on the couch are usually the nights I needed tracking most.
Patterns I Couldn’t See
Best recovery day: Sunday (avg 2h15m deep sleep). Worst: Friday (avg 1h05m deep). No surprise there — the week’s accumulated debt comes due.
Stress peaks at 11am-1pm and 6-8pm. The first is probably work pressure hitting its stride. The second might be the commute, or just the body’s natural circadian dip meeting whatever chaos the day delivered.
Stress lows at 7-9am. Morning routines working as designed.
The Prediction Problem
Today’s prediction based on last night (50m restorative, 22%): expected stress 35-45, likely climbing to 40+ by evening. Current reading at 23 suggests the storm hasn’t arrived yet.
This is where it gets useful. Not looking backward at what happened, but forward at what’s coming. The recommendation is simple: aim for 7.5h+ tonight, avoid alcohol and screens, sleep before 11pm.
My deep sleep improves 40% when I sleep before 11pm. Knowing that number makes the tradeoff concrete. An extra hour of Netflix costs me almost half my recovery.
Balance as a Practice
The five core features here aren’t revolutionary: restorative sleep ratio, sleep-to-stress correlations, day-of-week patterns, time-of-day patterns, and predictive alerts. What’s different is having them synthesized into something actionable instead of scattered across a dozen charts.
Health isn’t a dashboard problem. It’s a story problem. The data needs a narrator.
Take care of yourselves out there. Get some good rest tonight.
Related: When AI Replaces Dashboards - The earlier post about reverse-engineering Garmin’s API to understand recovery telemetry.