Simulating Circadian Rhythm Disruptions and Cognitive Performance on Astronauts Using Machine Learning Models
Abstract
Circadian rhythms govern essential physiological processes such as sleep-wake cycles, hormone regulation, and cognitive function. Disruptions to these rhythms, often caused by artificial light exposure or irregular sleep patterns, are a significant challenge for astronauts on long-duration space missions, where natural environmental cues are absent. This research aims to develop a machine learning-based approach to simulate and analyze circadian rhythm disruptions and their effects on cognitive performance. By leveraging time-series data, we model the influence of artificial light exposure on sleep patterns, alertness, and cognitive outcomes, using a combination of Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM) networks, and regression models. We will simulate various lighting conditions (e.g., blue light exposure) and examine their impact on key circadian biomarkers such as sleep onset, melatonin secretion, and cognitive performance metrics (reaction time, memory recall). This study will also incorporate biological data (e.g., melatonin and cortisol levels) to model more realistic circadian rhythms. In addition to synthetic data, we will utilize publicly available datasets on sleep and cognitive performance to train and validate the models. The final goal is to provide insights into how light exposure schedules could mitigate sleep disturbances and cognitive decline for astronauts, helping to develop countermeasures for deep-space missions. This machine learning framework represents a low-cost, accessible solution to simulate and study spaceflight-related circadian rhythm disruptions and their consequences on astronaut performance, offering new avenues for future research on astronaut health and performance optimization during extended space missions.