Ensuring a sustainable supply of food resources and effective atmospheric revitalization remains a critical challenge for long-duration space missions. This challenge is further complicated by the technical constraints and high costs associated with repeated in-orbit experimentation. In response, this study investigates and models the biological responses of Spirulina microalgae, a recognized protein-rich organism, under suborbital flight conditions, with the overarching goal of replacing conventional trial‑and‑error approaches with an intelligent predictive framework. To this end, experimental data capturing complex environmental stressors, including microgravity and distinct light wavelengths (red, white, and dark), were utilized to train an advanced neural network. Specifically, a Deep Q‑Learning (DQN) algorithm was developed, incorporating mechanisms such as experience replay to mitigate data fluctuations and enhance learning stability, and was trained over 5,000 episodes. The results demonstrate the model's outstanding performance, achieving a coefficient of determination (R²) exceeding 0.98 for all output indicators, while the root mean square error for light‑related parameters fell below 0.013, indicating an excellent agreement between predictions and actual observations. Although minor fluctuations were observed in the prediction of time‑dependent parameters, the overall convergence of the model confirms that artificial intelligence can serve as a reliable, cost‑effective, and efficient tool for the automated monitoring and optimization of biological responses in future space probes and orbital stations. This work thus represents a meaningful step toward ensuring astronaut biosafety in deep‑space exploration.