# Open Raw-Nightly Scout — 2026-03-28 ## Purpose The project's main bottleneck is now precise: - one cohort that combines - true `participant-night` structure, - dense repeated burden, - and strong nightly raw physiology. This note answers a narrower question: - does an **open** dataset currently visible from public sources plausibly clear that bar, or at least come close enough to reduce dependence on institutional acquisition? ## Current reading No open dataset identified in this scout currently closes the bottleneck. The best practical reading is: - **open direct-fit candidate:** none - **restricted but structurally promising:** `TILES-2018`, `Predicting chronic stress among healthy females...` - **restricted and off-lane:** `mcPHASES` - **open but off-lane:** `MMASH`, `Wearable Device Dataset from Induced Stress and Structured Exercise Sessions` ## Candidates ### 1) TILES-2018 Status: - `near-fit but access-restricted` Why it matters: - `n = 212` over `10` weeks - wearable sensors plus repeated well-being / behavioral surveys - explicit sleep and heart-rate coverage Why it does not close the bottleneck today: - it is not open-only in the current workflow - it remains an access/DUA path, not a plug-in public cohort Source: - - ### 2) Predicting chronic stress among healthy females using daily-life physiological and lifestyle features from wearable sensors Status: - `near-fit but restricted` Why it matters: - `129` participants - `7` consecutive days - minute-level rows with heart rate, sleep status, and lifestyle features - chronic stress framing is structurally relevant Why it does not close the bottleneck today: - files are restricted - follow-up depth is short - current access still depends on request approval Source: - ### 3) mcPHASES Status: - `restricted and off the primary stress lane` Why it matters: - recent PhysioNet release with very rich wearable and diary structure - daily diary plus many Fitbit-derived tables - includes sleep, heart rate, temperature, respiratory rate, and stress scores - participant timelines are explicit Why it does not close the bottleneck today: - restricted health-data access - the primary burden family is menstrual health / symptoms, not the current repeated stress lane - it may become useful for future raw-nightly non-stress or somatic work, but not as the clean stress closure Source: - - ### 4) MMASH Status: - `open but too small` Why it matters: - beat-to-beat and actigraphy are real - stress inventory and multiple psychological measures exist Why it does not close the bottleneck today: - only `22` participants - effectively `~1` night per participant in the usable nightly adaptation - useful as support or component validation only Source: - ### 5) Wearable Device Dataset from Induced Stress and Structured Exercise Sessions Status: - `open but event/session-only` Why it matters: - open - Empatica-derived physiology - explicit stress induction protocol Why it does not close the bottleneck today: - induced acute stress sessions, not participant-night longitudinal sleep - useful for event physiology or challenge work, not the nightly transport bottleneck Source: - ## Decision The open-only path is not currently enough to close the main raw-nightly bottleneck. The correct interpretation is: 1. the internal repo is exhausted for this lane 2. the open web scout does not surface a clean direct-fit cohort 3. the shortest path to closure is still: - restricted/credentialed access to a strong cohort - or institutional acquisition ## Correct next move Do not spend another cycle trying to rescue: - tiny support cohorts, - acute session datasets, - or symptom-rich but off-lane datasets into the current `raw nightly stress` lane. Keep the bar where it is: - dense repeated burden - strong raw nightly support - participant-night linkage and treat `TILES-2018` plus institution-grade acquisition as the primary closure paths.