THE SCIENCE
Peer-reviewed methodology, not marketing claims
Every model inside RosterWise has a published paper behind it. Every claim on this site has a citation. We tell you what we know — and what we do not.
ACADEMIC PARTNERSHIPS
Developed with two of the UK's leading sleep and occupational-health research centres
Oxford University SCNi
Sleep and Circadian Neuroscience Institute
One of the world's leading centres for sleep and circadian research, directed by Professor Russell Foster CBE FRS. The Institute's research on circadian biology and sleep regulation underpins the predictive models inside RosterWise.
King's College London
Occupational health and shift-work epidemiology
Contributing expertise in occupational health and the epidemiology of shift work. KCL's longitudinal research on shift patterns and worker outcomes informs how RosterWise translates model output into operational guidance.
These partnerships ensure that RosterWise's models reflect the current scientific consensus, not legacy assumptions.
SCIENTIFIC FOUNDATION
Three peer-reviewed pillars, reconciled into one engine
MODEL 01
The Three-Process Model of Alertness
Originally published by Åkerstedt and Folkard (1997) and refined by Ingre et al. (2014), this model predicts alertness from the interaction of homeostatic sleep pressure (Process S), circadian rhythm (Process C), and sleep inertia (Process W). It is the most widely validated biomathematical fatigue model in European occupational health research.
Åkerstedt & Folkard, 1997 · Ingre et al., 2014
MODEL 02
The HSE Fatigue Risk Index
Developed by Spencer, Robertson and Folkard for the UK Health and Safety Executive (Research Report RR446, 2006). This peer-reviewed methodology quantifies the relative risk of a shift pattern based on timing, duration, rest periods, and cumulative work. RosterWise implements the published methodology as one layer of a multi-model approach.
Spencer, Robertson & Folkard, 2006 (HSE RR446)
MODEL 03
Probabilistic Sleep Estimation
Drawing on population-level sleep data from shift-worker studies, RosterWise estimates likely sleep timing and duration when direct measurement is unavailable, using Bayesian inference calibrated by shift type, commute duration, and chronotype. The approach is grounded in the published shift-worker sleep literature.
Fischer et al., 2017 · Åkerstedt et al., 2004
WHY FOUR MODELS
A single model can be gamed, misapplied, or stretched past its validated range
This is precisely what happened with the HSE Fatigue Risk Index. Employers used threshold scores as pass/fail criteria for individual shifts — a purpose for which the tool was never designed or validated. The HSE withdrew the tool in 2021 partly because of this misuse pattern.
RosterWise prevents this by running four independent models in parallel and reconciling their outputs through a Risk & Intervention Classifier. Where models agree, confidence is high. Where they disagree, the system flags uncertainty rather than guessing.
VALIDATION APPROACH
A three-stage strategy, applied to every model layer
01
Internal consistency
Cross-model agreement rates and sensitivity analysis against published benchmark shift patterns. Where models agree, confidence is high. Where they disagree, the system flags uncertainty rather than guessing.
02
External benchmarking
Comparison of RosterWise outputs against published fatigue risk assessments for standard shift patterns, including the HSE's own benchmark schedules from the RR446 report.
03
Field validation
Planned prospective studies with partner organisations, correlating RosterWise predictions with operational safety data, incident reports, and worker-reported alertness.
REFERENCES
Key publications
- 01Åkerstedt, T., & Folkard, S. (1997). The three-process model of alertness and its extension to performance, sleep latency, and sleep length. Chronobiology International, 14(2), 115–123.
- 02Åkerstedt, T., Kecklund, G., & Johansson, S.-E. (2004). Shift work and mortality. Chronobiology International, 21(6), 1055–1061.
- 03Fischer, D., Lombardi, D. A., Folkard, S., Willetts, J., & Christiani, D. C. (2017). Updating the 'Risk Index': A systematic review and meta-analysis of occupational injuries and work time-related risk factors. Chronobiology International, 34(10), 1423–1438.
- 04Ingre, M., Van Leeuwen, W., Klemets, T., Ullvetter, C., Hough, S., Kecklund, G., Karlsson, D., & Åkerstedt, T. (2014). Validating and extending the three-process model of alertness in airline operations. PLOS ONE, 9(10), e108679.
- 05Spencer, M. B., Robertson, K. A., & Folkard, S. (2006). The development of a fatigue / risk index for shiftworkers. HSE Research Report RR446. Health and Safety Executive.
A formatted reference list of the full peer-reviewed foundation is available on request for procurement and academic reviewers.