The　digital　economy　is　a　new　type　of　economic　form　with　new　expression　concepts,　many　influencing　factors,　and　a　wide　scope.　Accurate　prediction　of　it　can　not　only　make　quantitative　judgments　on　future　economic　trends　but　also　directly　serve　the　development　of　the　economy　and　society　and　effectively　prevent　economic　risks.　This　paper　combines　the　Pearson　correlation　coefficient　and　the　idea　of　feature　importance　assessment　of　random　forests　to　construct　an　index　system　for　predicting　the　impact　factors　of　China＇s　digital　economy.　And　based　on　the　QPSO　algorithm　and　LSTM　model　to　construct　China＇s　digital　economy　scale　forecasting　model,　the　validity　of　the　model　is　empirically　analyzed　by　China＇s　digital　economy　scale　data　from　1993　to　2020.　The　results　show　that　the　idea　of　combining　a　metaheuristic　algorithm　and　neural　network　has　validity　in　China＇s　digital　economy　scale　forecasting　work.　In　addition,　the　QPSO-LSTM　model　outperforms　both　the　single　LSTM　model　and　the　PSO-LSTM　model　in　terms　of　parameter　search　results,　convergence　speed,　MAPE,　MAE,　and　R2　evaluation　indexes.　©　2023　Technical　Committee　on　Control　Theory,　Chinese　Association　of　Automation.