Sleep regularity and duration across the internship window.
Sleep was recorded on 68 of 68 eligible internship mornings (100% coverage). Median sleep was 6.1 hours; 36 nights were over six hours and 9 were under four hours.
Compared with all available pre-internship records, available data suggest sleep duration was 0.7 hours lower (consistent evidence), recovery was 12 points lower (consistent evidence), HRV was 12 ms lower (consistent evidence), and resting heart rate was 7 bpm higher (consistent evidence).
Following mornings after the listed social nights were associated with 1.3 more hours of sleep and 15 points lower recovery in weekday-matched comparisons (matched n≥18; evidence varies by metric).
Each cutoff is a descriptive threshold, not a medical diagnosis or personalized target.
Internship began June 1 · Health through Aug 7 · NYC events through Aug 9
Social events are plotted on the following morning. Filled event bubbles sit on recorded recovery; hollow bubbles above the plot indicate no recovery value.
Hover a point to inspect its day and values. Keyboard: focus a chart and use the left and right arrow keys.
Changes show the first cohort minus the second.
Internship first All prior data second
| Metric | Cohorts · first vs second | Change | Evidence |
|---|---|---|---|
| Sleep duration | Mean5.9h vs 6.7hMedianmedian 6.1h vs 6.6hCoverage68 of 68 (100%) vs 268 of 275 (97%) | −0.7 hMedian Δ −0.5 h | Consistent association |
| Sleep performance | Mean55% vs 65%Medianmedian 60% vs 72%Coverage68 of 68 (100%) vs 268 of 275 (97%) | −9 ppMedian Δ −13 pp | Consistent association |
| Recovery | Mean42% vs 54%Medianmedian 40% vs 59%Coverage68 of 68 (100%) vs 268 of 275 (97%) | −12 ppMedian Δ −19 pp | Consistent association |
| HRV | Mean29ms vs 41msMedianmedian 29ms vs 42msCoverage68 of 68 (100%) vs 268 of 275 (97%) | −12 msMedian Δ −12 ms | Consistent association |
| Resting heart rate | Mean73bpm vs 66bpmMedianmedian 71bpm vs 64bpmCoverage68 of 68 (100%) vs 268 of 275 (97%) | +7 bpmMedian Δ +7 bpm | Consistent association |
Internship comparisons use each metric’s full available pre-internship history from the same selected source. Positive changes mean the first cohort was higher. Social-night rows also show the weekday-matched estimate when available. Matched weekday sample sizes: n=18.