Moon Missions and Modeling
- Jaime K. Devine, PhD
- 10 hours ago
- 6 min read
We are all rightfully excited about NASA’s recent mission to the moon. Artemis II lifted off from NASA’s Kennedy Space Center on Wednesday, April 1, sending NASA astronauts Commander Reid Wiseman, Pilot Victor Glover, Mission Specialist Christina Koch, and Canadian Space Agency astronaut Mission Specialist Jeremy Hansen on a 10-day test flight around the Moon and back. On Monday, April 6, the Artemis II crew reached the mission's maximum distance from Earth at 252,756 miles (406,771 km), beating the previous record set by the Apollo 13 mission in 1970 by 4,111 miles (6,616 km). Artemis II also made history by sending the first woman, first person of color, and first Canadian astronaut to the Moon. The crew of Artemis II splashed down safely off the coast of San Diego on the evening of Friday, April 10, 2026 [1-3].
Fabulous news, but what does this historic mission have to do with fatigue risk management and biomathematical modeling? I’m so glad you asked! Unsurprisingly, the launch itself is just the tip of the iceberg with regard to the amount of planning that went into making the mission a success. NASA scientists have been hard at work for years preparing for Artemis II and its proposed follow-up missions. And yes, that work included studying the kinds of sleep disruptions and performance impairments crew members could expect using mission simulations and biomathematical models.
You might think that nothing would be more conducive to sleep than a zero-gravity bedroom thousands of miles away from the problems of Earth but sleeping in space isn’t easy. Even when they are given eight hours of sleep opportunity, astronauts only achieve about six hours of sleep. One reason for short sleep is the lack of regular light-dark cycles and the heavy use of artificial light aboard ship. Uncomfortable sleeping conditions (like having to strap yourself down so you don’t float away), background noise from the ship and fellow crew members, the disorienting effects of no gravity, and the stress and excitement of being in space can all contribute to sleep disturbances in astronauts who may be ideal sleepers back on Earth.
Space exploration relies on small crews being able to maintain a high level of performance with limited support from mission control. There is a serious risk of operational errors occurring if astronauts are cognitively impaired due to sleep debt. As a harrowing example, fatigue is frequently cited as one of the contributing factors leading to the 1986 Challenger STS-51L Accident that resulted in the death of seven astronauts[4]. One of the ways that NASA prepared for the Artemis missions was to conduct a 45-day simulated space mission that included chronic sleep restriction in order to characterize any changes in performance using the psychomotor vigilance task (PVT) and subjective fatigue ratings[5]. The study was led by researchers at the NASA Ames Research Center Fatigue Countermeasures Laboratory (view article) and conducted in the Human Analog Research Exploration (HERA) habitat (view article). The results of the NASA simulation were originally published in the journal Scientific Reports in September 2020, well before the April 2026 Artemis launch. The full article is available here.
Participants in this simulation needed to be “astronaut-like.” That meant that eligible participants had to be non-smokers, 30–55 years old, proficient in the English language, and had to have at least a Master of Science in a science, technology, engineering, mathematics (STEM) discipline or the equivalent years of experience. Participants had to meet NASA long-duration spaceflight physical standards verified by a physical exam. Volunteers who passed all these requirements were split into five separate crews of four crew members each, just like for the real mission. If you wanted to compare the study participation requirements to the actual credentials of the Artemis II crew, check out the team’s bios here.
Each simulation followed the same 45-day schedule, which included five hours of sleep for five nights in a row followed by eight hours of sleep for two nights in a row on a repeating cycle. Participants were monitored continuously by simulated mission control and performed tasks similar to what an astronaut would be expected to do in space. Participants also completed a five-minute version of the PVT and a Samn-Perelli subjective fatigue rating scale five times a day approximately every three days during the simulation. The study found that objective PVT performance, but not Samn-Perelli ratings, declined from the beginning to the end of the mission. This disconnect between subjective and objective fatigue was not a shocking result; it is well-documented across laboratory and field studies. However, it still represents a potential hazard if astronauts were to rely only on self-assessment of fatigue during a mission.
Taking a note from commercial aviation, NASA evaluated how well biomathematical models predicted performance changes due to sleep loss compared to objective performance during the simulation. Four models were tested against the actual PVT data, including the Adenosine-Circadian model, the Unified Model of Performance (UMP), the Washington State University (WSU) State-space model, and the Sleep, Activity, Fatigue, and Task Effectiveness (SAFTE) model[5].
The biomathematical models were able to predict average changes across the mission. Figure A below demonstrates the relationship between model predictions and actual PVT lapses by day of the mission. Raw PVT lapse scores are shown as filled circles, with black circles indicating days following 5 hours of sleep and red circles indicating days following 8 hours of sleep. Model predictions are shown as open symbols as follows: triangles = adenosine-circadian model, squares = UMP model, diamonds = WSU model, stars = SAFTE model. SAFTE Effectiveness is plotted in the inverse on the left y-axis. The raw model lapse predictions were higher than the observed group-average lapse values throughout the mission. Rescaled models were all able to predict lapses by mission day on average, as shown in Figure B.
All the models in the NASA simulation did a poorer job at predicting individual PVT performance than they did population averages because resilience to sleep loss varied greatly across the participants. Some individuals were able to sustain performance in the face of sleep restriction better than others, despite the fact that all of the participants were rigorously selected to be “astronaut-like”. This means that being young, healthy, physically fit, highly-motivated, and well-educated does not automatically make you resilient to the effects of sleep loss. Participants in the simulated spaceflight could not accurately track their increased fatigue risk either. Taken together, these findings highlight the risk of relying on the human ego to accurately monitor fatigue. Models, in comparison, examine fatigue risk at the population level and do not currently account for individual differences. The benefit of this kind of generic risk assessment means that everyone is being treated equally, whether they’ve been to the moon or just followed along on YouTube. Generic risk assessment can be used to create schedules that can apply to a wide range of workers rather than just one specific individual.
The NASA study concluded that biomathematical modeling would be a useful tool for developing crew schedules. In January 2026, NASA posted an Artemis II Overview Timeline that clearly shows time blocked out for sleep and pre/post sleep periods. We use a similar blocking for pre/post sleep periods called “Prep and Unwind” in SAFTE-FAST. The Overview Timeline does not report whether any biomathematical models were used to create the crew schedules. Also not included in the timeline is the playlist of songs that mission control used to wake up the astronauts during the mission. Fortunately, you can find the playlist on Spotify.
According to the Artemis II Reference Guide, astronauts wore wristband devices (actigraphy) to monitor sleep and activity throughout the mission. The guide mentions that the actigraphy data will be used for safety assessments postflight evaluations of cognition and sleep quality, but does not mention any plans for retrospective biomathematical modeling. Hopefully, we will see a follow-up to the 2020 NASA biomathematical modeling paper that reports how well SAFTE and other biomathematical models predict performance based on actual data collected in space. NASA scientists will have to work quickly if they are hoping to implement any changes to astronauts’ sleep schedules based on feedback from the Artemis II mission. The next Artemis mission is scheduled for launch as early as 2027!
References
Comments