Photosynthesis rate and transpiration rate are crucial physiological indicators in plants. In a controlled artificial environment, Italian lettuce was chosen as the research subject. A nested experiment was conducted to investigate the multivariate impact on the photosynthesis rate and transpiration rate of lettuce. The study unveiled patterns of environmental factors affecting these rates, leading to the construction of a neural network prediction model for photosynthesis rate and transpiration rate during the seedling phase of lettuce. For lettuce seedlings, four factors were selected: temperature, relative humidity, photosynthetic photon flux density (PPFD), and environmental CO2 concentration. Using the random forest method, a correlation analysis of the data was carried out. The results revealed that factors strongly correlated with the transpiration rate, in descending order, were CO2 concentration, temperature, relative humidity, and PPFD. Meanwhile, for the photosynthesis rate, the factors were CO2 concentration, PPFD, temperature, and relative humidity. A GA-BP neural network physiological indicator prediction model was developed, employing the enumeration method to determine the number of hidden layer nodes and training functions, and optimizing the initial weights and thresholds of the BP neural network through a genetic algorithm. Testing with actual data, the determination coefficients of predicted and actual values for photosynthesis rate and transpiration rate were 0.96212 and 0.97944, respectively, with root mean square errors (RMSE) of 2.9832μmol/(m2·s) and 0.0014358mol/(m2·s). This demonstrated the significantly improved performance of the GA-BP neural network in terms of model accuracy and iteration times. In summary, the research result can provide a valuable basis for environmental regulation in facility lettuce production.