Application of deep learning in assessing the impact of industrial digitalization and transformation on regional economic development

Industrial digitalization and transformation are vital drivers of regional economic growth, promoting innovation and sustainability; however, accurately evaluating the complex, nonlinear effects is limited by fragmented data, regional disparities, and the low interpretability of current analytical models. This research investigates the use of deep learning (DL) techniques to assess the impact of digital transformation on regional economic growth, sustainability, and productivity enhancement. Using an Industrial Digitalization & Regional Growth dataset with 2100 rows spanning from 2019 to 2024 across various Chinese regions, the research integrates Social Network Analysis (SNA) and DL models to identify and evaluate the key drivers influencing regional economic performance. Digital Total Factor Productivity (DTFP) is employed as the primary indicator to measure high-quality economic development. The data was preprocessed through text cleaning, tokenization and Lemmatization. The features were retrieved through Term Frequency-Inverse Document Frequency (TF-IDF) and Bidirectional Encoder Representations from Transformers (BERT). Industrial automation contributes positively to DTFP, reflecting the productivity-enhancing effects of digitalization, whereas EPU shows a negative influence, indicating its constraining role in sustainable growth. Among multiple predictive models, the Magnetotactic Bacteria Optimized Spatio-Temporal Graph neural network (MBO-ST-GNN) exhibits improved performance MAE = 0.2984, MSE = 0.0896, RMSE = 0.2993, and R 2 = 0.9875 in capturing temporal dependencies and accurately forecasting DTFP trends. Overall, the findings provide valuable theoretical and practical guidance for policymakers, enabling data-informed strategies to foster digital infrastructure, enhance industrial competitiveness, and promote balanced, green, and sustainable regional development.