PerfBoost-Net: A ResNet-Capsule Framework for Analyzing the Impact of AI Tools on Construction Workforce Productivity and Satisfaction

The rapid integration of artificial intelligence (AI) tools in construction offers significant potential to improve workforce productivity and satisfaction. This study introduces PerfBoost-Net, a novel framework combining ResNet and Capsule Networks to analyze the impact of AI tools on construction workforce metrics. The model leverages advanced deep learning techniques to predict and interpret productivity and satisfaction metrics, utilizing datasets enriched with features such as tool adoption rates, task complexity, and safety perceptions. PerfBoost-Net achieved a prediction accuracy of 97.2% for productivity and 94.6% for satisfaction, outperforming baseline models like standalone ResNet and Capsule Networks. Feature importance analysis, conducted using SHAP, revealed tool adoption rate and task complexity as the most influential factors. The framework also categorized AI tools into automation, analytics, and safety enhancement, showcasing their distinct impacts, with automation tools contributing to a 45% improvement in productivity and safety tools increasing satisfaction by 50%. The proposed model provides actionable insights for optimizing AI tool deployment in the construction industry. The results validate PerfBoost-Net as an effective decision-support tool, offering the potential for scalability to other domains. Future work aims to enhance the model's interpretability and extend its application to diverse construction environments.