Abstract: To reveal the spatiotemporal correlation patterns of earthquake occurrence frequency in the Longmenshan Fault System in western Sichuan, this study systematically collected the earthquake catalog from 1970 to 2025. The minimum completeness magnitude Mc=3.1 was determined based on the Gutenberg-Richter (G-R) relationship, and nine time series of earthquake occurrence frequency were constructed in terms of latitude and magnitude. The Cross Correlation Function (CCF) was adopted to mine the correlation characteristics. The core findings indicate that seismic activities in the central and northeastern segments of the Longmenshan Fault show a synchronous response, while the southwestern segment presents a significant 5-year lag in earthquake occurrence relative to the central and northeastern segments. No time lag exists between high-magnitude and low-magnitude earthquakes within each segment. To verify this pattern and achieve accurate prediction, the 5-year lag feature was taken as the model input. A hybrid GA-LSTM prediction model was established by introducing the Genetic Algorithm (GA) to optimize the hyperparameters of the Long Short-Term Memory (LSTM) network. Experimental results show that the proposed model improves the prediction accuracy by approximately 88% compared with the baseline LSTM and ARIMA models, and can effectively characterize the nonlinear lag correlation of the fault zone. Based on the optimal model, the seismic activity in the southwestern segment from 2026 to 2030 was predicted. The results demonstrate that the region is dominated by low-magnitude earthquakes, with high-magnitude earthquakes concentrated in 2027—2029, and the overall seismic situation is stable. The core contribution of this paper is the quantitative identification of the lag pattern of seismic activity in the Longmenshan Fault Zone. All analyses and conclusions in this paper are based on the statistics of annual earthquake occurrence frequency, and only reflect the temporal correlation characteristics of seismic activity in the fault system. Under this premise,this paper quantitatively identifies the temporal lag pattern of seismic activity in the Longmenshan Fault Zone. The established model can provide scientific support for regional seismic hazard assessment and disaster prevention and mitigation, and also offer technical references for the activity analysis of large-scale fault systems.