Related Experiment Video
Updated: May 25, 2026
08:36
Collecting Sleep, Circadian, Fatigue, and Performance Data in Complex Operational Environments
Published on: August 8, 2019
Predicting pilot's sleep during layovers using their own behaviour or data from colleagues: implications for
Jillian Dorrian1, David Darwent, Drew Dawson
1Centre for Sleep Research, School of Psychology, Social Work and Social Policy, University of South Australia, GPO Box 2471, Adelaide, South Australia 5001, Australia. jill.dorrian@unisa.edu.au
Accident; Analysis and Prevention
|January 14, 2012
Summary
Predicting pilot sleep patterns using individual data showed modest improvements over using colleague data. Personalized sleep predictions offer slight benefits for understanding rest in demanding work environments.
Area of Science:
- Occupational Health
- Sleep Science
- Biomathematics
Background:
- Biomathematical models predict work performance and safety based on sleep patterns.
- Individual responses to sleep loss vary, necessitating personalized prediction models.
- Past behavior may predict future responses to similar work conditions.
Purpose of the Study:
- To investigate the predictive value of individual sleep timing and duration data.
- To compare individual sleep prediction accuracy against colleague and random samples.
- To assess the utility of personalized data in biomathematical sleep models for pilots.
Main Methods:
- Collected sleep diaries and wrist actigraphy from 306 international long-haul pilots over at least two weeks.
- Analyzed sleep and wake patterns, and total sleep time (TST) for 50 equivalent layovers completed twice by individual pilots.
- Compared individual sleep prediction accuracy against data from colleagues (n=2311) and random samples.
Main Results:
- Using an individual's own sleep data improved prediction concordance by approximately 5% over a large sample of different pilots and 10% over a random sample.
- Individual TST prediction yielded r=0.83, compared to r=0.78 for colleague data and r=0.73 for random samples.
- Mean TST difference was <20 min using individual data versus <40 min using colleague data, though confidence intervals were large.
Conclusions:
- Personalized sleep data offers a modest improvement in predicting sleep behavior for international pilots on specific layover patterns.
- Individual past behavior provides only a slight advantage over colleague sleep data for predicting future sleep.
- Further research may refine the application of individual sleep data in biomathematical models for occupational safety and efficiency.
