Cover illustration for Resolving Luteal Sleep Fragmentation With One Unified Model

What the AI found

“Luteal phase sleep disturbances clustered exclusively on days with late afternoon caffeine, while follicular phase sleep remained unaffected.”

Before

four disconnected apps, zero shared timeline

After

one unified table she reviews in three minutes

The same system, three states — real screens, not a screenshot

1Starting
Apple Notes
  • Oura: sleep score 61, elevated temp (+0.3C)
  • Clue: Day 21, luteal, foggy morning
  • MyFitnessPal: 2 espressos, 1 matcha afternoon
  • Notes: woke at 03:15 again with racing pulse
2Working
Claude

Prompt

Cross-reference my cycle days, afternoon caffeine logs, and nocturnal waking events from the last 90 days. Is the luteal sleep disruption uniform?

Cross-reference my cycle days, afternoon caffeine logs, and nocturnal waking events from the last 90 days. Is the luteal sleep disruption uniform?

AI

No. Across 90 days, your luteal waking events clustered exclusively on days with caffeine intake after 14:00 (8 of 9 poor nights). On luteal days with zero caffeine after 13:30, sleep efficiency matched your follicular baseline (89% vs 91%). Follicular sleep showed no sensitivity to 16:00 tea.
3Implemented
Apple Health

+28 min baseline

Luteal REM stability

14:00 strict

Afternoon tea cut-off

28 of 28 days

Log consistency

IndividualDone-for-you in use

Resolving Luteal Sleep Fragmentation With One Unified Model

How a done-for-you data synthesis untangled caffeine sensitivity from cycle phases.

A 36-year-old product strategist, Northern Europe

Tools used

The real tools used here — swap any for your own equivalent. Each links to how we’d set it up.

4 min readWellness & AI editorial
1

Before anything was set up

Her bedside table held two notebooks; her phone stored four specialised health apps that never communicated. Every month around day twenty, sleep degraded, resting heart rate climbed, and morning focus vanished. She had spent eighteen months assuming luteal phase restlessness was an unalterable biological fact, logging symptoms faithfully across separate silos yet unable to extract an actionable signal from ninety days of fragmented data.

Apple Notes
  • Oura: sleep score 61, elevated temp (+0.3C)
  • Clue: Day 21, luteal, foggy morning
  • MyFitnessPal: 2 espressos, 1 matcha afternoon
  • Notes: woke at 03:15 again with racing pulse
2

Done-for-you, doing its job

We unified ninety days of exported CSVs into a single clean relational table in Google Sheets, then configured Claude to evaluate interactions between cycle timing, resting heart rate spikes, and stimulant intake. Instead of generic advice, the system examined her personal data across three complete cycles, isolating a sharp metabolic divergence she had completely missed.

Claude

Prompt

Cross-reference my cycle days, afternoon caffeine logs, and nocturnal waking events from the last 90 days. Is the luteal sleep disruption uniform?

Cross-reference my cycle days, afternoon caffeine logs, and nocturnal waking events from the last 90 days. Is the luteal sleep disruption uniform?

AI

No. Across 90 days, your luteal waking events clustered exclusively on days with caffeine intake after 14:00 (8 of 9 poor nights). On luteal days with zero caffeine after 13:30, sleep efficiency matched your follicular baseline (89% vs 91%). Follicular sleep showed no sensitivity to 16:00 tea.
3

The finished system, running on its own

The handed-over build now runs autonomously. Every morning, wearable data syncs silently into her private spreadsheet. When her cycle enters the luteal phase, an automated calendar rule gently shifts afternoon hot drinks to rooibos after 14:00. She spends three minutes every Sunday scanning one consolidated summary before closing her laptop. The four separate tracking apps were deleted three weeks ago; the operational clarity remains entirely hers.

Apple Health

+28 min baseline

Luteal REM stability

14:00 strict

Afternoon tea cut-off

28 of 28 days

Log consistency

4 down to 1

Apps consolidated

3 minutes

Weekly review time

ClaudePattern synthesis & analysis

Large context window handles multimonth multimodal exports without truncating daily notes.

Google SheetsCentralised data repository

Simple, permanent ownership of normalised data across cycle markers, sleep, and habits.

Apple HealthPassive metric aggregation

Reliable local export of resting heart rate, body temperature shifts, and sleep duration.

These are the tools used in this story. Any can be swapped for an equivalent you already trust.

Commission your build

This story runs on Done-for-you. The tools and prompts above are the real build — swap any tool for your own equivalent and follow the same steps.

Suggested for you

Based on what you've been reading — always learning.

See all →