Sleepgenic / Research method · Version 2.0 · 28 September 2026

How we learn from repeated nights.

We connect a documented day to the next morning’s sleep outcome. The current series adds meals and final caloric timing to activity and wearable signals. The historical sleep-only record remains a separate baseline.

Research question

Which food, timing and movement patterns repeatedly precede changes in sleep duration, regularity and overnight recovery within a person, after the available context is considered? We do not treat a single night or wearable score as the answer.

Two records with distinct roles

• Active response log: one dinner date and following sleep date per row. Food photo or report, documentation quality, activity modality and dose, final caloric cutoff, reported bedtime, snack and alcohol flags, and next-morning Garmin metrics. The 28 September source has 22 linked observations; 21 include detailed sleep and recovery metrics, 21 have a Sleep Score, and 20 have a valid reported bedtime.
• Historical sleep record: Sleep Data workbook with 182 nights containing recorded duration across a November 2025–August 2026 calendar range. It helps define personal background variation and historical context. Because meal records are absent, it cannot answer the new food-response question.

Both sources retain original dates, units, provenance and missing values. The active log is the sole continuing daily research input. We do not pool historical and active sleep scores without checking device, metric and context consistency.

Observation and time linkage

Day D holds the meal and activity exposure. Day D+1 is the next morning’s sleep result. The final caloric cutoff means the last intake containing calories, even if dinner ended earlier. Bedtime is the reported time of going to bed, not verified sleep onset. A time after midnight belongs to the correct sleep window. Compute final-intake-to-bed intervals only when both times are present and the date rollover is unambiguous; otherwise label the interval unavailable. Preserve source timestamps and distinguish any GMT times in the older workbook from reported local times.

Outcomes and measurement

Prioritize total sleep time and, once reliable timing is available, sleep regularity. Examine HRV, resting and overnight heart rate, awake time and Body Battery as related but separate recovery measures. Sleep Score summarizes the device’s model; deep and REM stages are exploratory wearable estimates. Do not combine these into a new precision score. Note device and firmware changes when known and keep definitions stable across comparisons.

Analysis procedure

1. Audit source dates, paired-night counts, missingness, uncertain photos and corrections. Start with the full chronology, then the newest observation.
2. State the proposed exposure and primary outcome before comparing nights. Use comparable meal types, timing windows and activity doses when available; show the included observation IDs and both favorable and conflicting results.
3. Check bedtime, prior sleep, exertion, alcohol, illness, travel, heat and other recorded context. A field that was not captured is unknown, not a negative answer.
4. Separate observed association from a planned test. If feasible, use repeated, prespecified within-person comparisons with stable measurement and sensible exposure choices. Do not recommend risky exposures to manufacture variation.
5. Record the result, competing explanations, confidence status, and the next observation that would distinguish them. Revise the public evidence register when a finding changes.

Publication thresholds

Label each statement single observation, repeated association, tested within-person result, or unresolved. Replication across other consenting participants is needed before suggesting a generalizable response. For every release include source workbook/version, date window, available-night denominator, method, counterexamples and changes from the previous release. Preserve older reports as dated snapshots and correct confirmed transcription errors visibly.

Limits and safety

Meals are partly photo-inferred; exact nutrients and portions are usually unknown. Activity, meal composition and bedtime can change together. Wearable stages are estimates. A personal association does not prove causation or a longevity outcome. This research is educational and does not diagnose or treat insomnia or sleep-disordered breathing. Clinical concerns require an appropriate clinician.