Methodology
How Shiftwell computes scores (MVP version) and how it relates to the scientific basis.
Scores (MVP v0.1)
- Risk score (SLI proxy): weekly load/risk score derived from work/sleep schedule.
- Sleep score (proxy): combines average sleep duration + regularity proxy.
- Adaptability score (main): combines inverse risk + sleep score (proxy v0.1).
Scientific source
The scientific basis comes from a chronobiology/public health paper. The MVP currently implements a proxy logic for a fast product pilot, then will be aligned more strictly with the protocol.
Paper ↔ Shiftwell mapping
| Article item | Shiftwell variable | Status |
|---|---|---|
| Hours worked | Computed from weekly work segments | Implemented |
| Long shifts | Count of long work segments (proxy threshold) | Implemented (proxy) |
| Recovery | Longest recovery window between shifts | Implemented (proxy) |
| Night shifts | Overlap with biological night window | Implemented (proxy) |
| Biological loss | Work encroachment in biological window (23:00–07:00 proxy) | Implemented (proxy) |
| Social loss | Work encroachment in social window (proxy) | Implemented (proxy) |
| SRI/TST | Regularity proxy + total sleep time | MVP proxy |
Current limitations
- Current version = proxy v0.1 (thresholds/weights still to be aligned with final protocol).
- Self-reported input (no actigraphy).
- Adaptability score not cohort-calibrated in V1.
- No clinical recommendations in output (intentional).
Method roadmap
- Strict SLI thresholds alignment with paper/protocol
- Add average reference (paper or cohort baseline)
- Calibrate adaptability score on cohort data
- Move to a true visual calendar grid (drag/drop) if needed