In order to explore a more accurate method for predicting settlement in vacuum preloading foundation treatment, a vacuum preloading settlement prediction model based on Long Short-Term Memory (LSTM) neural network was constructed, using the example of the secondphase land reclamation project in the East Park of Xiamen New Airport planning area. Measured settlement data from two regions were selected as the dataset, and the results were compared with traditional settlement prediction methods (Asaoka Method, Three-Point Method, and Hyperbolic Method). The results show that the prediction model of LSTM neural network considering only sedimentation time series is more accurate than the traditional method considering only sedimentation time series. When vacuum film is damaged and settlement rebound occurs under the condition of vacuum precompression foundation treatment, RMSE and MAE of LSTM both decrease by more than 45% compared with the traditional method, and the prediction results of this method show a significant upward trend, which predict the settlement rebound situation accurately. In terms of prediction error, RMSE and MAE of the LSTM model which consider vacuum degree and sedimentation is more than 60% lower than those of the LSTM model which only consider sedimentation time series. This study can offer an advanced data-driven prediction method for vacuum preloading foundation settlement prediction.