Dynamic ensemble prediction of cognitive performance in spaceflight.

Area of Science:

• Space medicine
• Human factors engineering
• Cognitive neuroscience

Background:

• Spaceflight presents unique stressors like microgravity and isolation.
• These stressors can impair astronaut sleep, alertness, and performance.
• Neurobehavioral performance is crucial for mission success.

Purpose of the Study:

• To predict astronaut neurobehavioral performance during long-duration space missions.
• To identify key environmental, operational, and psychological factors influencing performance.
• To develop a robust predictive model for human behavior in extreme environments.

Main Methods:

• Longitudinal data from 24 astronauts on a 6-month International Space Station (ISS) mission.
• Repeated assessments of neurobehavioral performance using the Psychomotor Vigilance Test (PVT-B).
• An ensemble prediction model combining linear mixed effects, random forest, and functional concurrent models.

Main Results:

• The ensemble model significantly outperformed individual component models.
• Key predictors of PVT-B performance included prior performance, fatigue, stress, temperature, and radiation dose.
• The model accurately predicted individualized neurobehavioral performance.

Conclusions:

• An ensemble modeling approach effectively predicts astronaut neurobehavioral performance.
• Environmental and psychological factors significantly impact spaceflight performance.
• This predictive method is applicable to various human behavior prediction scenarios.