This study presents an interpretable machine learning (ML) approach to evaluate the effect of heat treatment (HT) processes on the creep performance of superalloys and predict creep rupture life with high accuracy. The method integrates classification and regression models with domain-specific knowledge, enhancing model interpretability and prediction accuracy. The HT process is identified as the most important descriptor, and the approach is validated using Waspaloy alloy, resulting in a 5.5-fold improvement in creep performance compared to previous studies. The ML framework includes three main steps: evaluating HT via domain-specific knowledge, evaluating HT via physics-informed classification models, and predicting creep rupture life using regression models with key feature screening. The study demonstrates that incorporating physical features significantly improves model performance, with the coefficient of determination (R²) increasing from 0.612 to 0.857. The approach also highlights the importance of key features such as grain size, γ′/γ′′ phase size, and cooling rate in determining creep performance. The study shows that the ML method can effectively optimize HT processes, leading to enhanced creep resistance and performance. The results indicate that the inclusion of physical features is crucial for improving the accuracy and interpretability of ML models in predicting creep rupture life. The study provides novel insights for enhancing the precision of predicting creep rupture life in superalloys and offers auxiliary support for the application of ML in the design of HT processes. The approach is validated through experimental data, demonstrating its effectiveness in optimizing HT processes and improving creep performance. The study also discusses the role of new features in classification models, the limitations of conventional SVR models, and the effectiveness of key feature screening in improving model accuracy. The results highlight the importance of HT parameters such as StT, AT, and PdT in determining creep rupture life and provide a framework for optimizing HT processes in superalloys. The study concludes that the proposed ML approach offers an efficient and effective means to evaluate the influence of HT processes on the creep properties of superalloys while significantly improving the accuracy of creep rupture life prediction.This study presents an interpretable machine learning (ML) approach to evaluate the effect of heat treatment (HT) processes on the creep performance of superalloys and predict creep rupture life with high accuracy. The method integrates classification and regression models with domain-specific knowledge, enhancing model interpretability and prediction accuracy. The HT process is identified as the most important descriptor, and the approach is validated using Waspaloy alloy, resulting in a 5.5-fold improvement in creep performance compared to previous studies. The ML framework includes three main steps: evaluating HT via domain-specific knowledge, evaluating HT via physics-informed classification models, and predicting creep rupture life using regression models with key feature screening. The study demonstrates that incorporating physical features significantly improves model performance, with the coefficient of determination (R²) increasing from 0.612 to 0.857. The approach also highlights the importance of key features such as grain size, γ′/γ′′ phase size, and cooling rate in determining creep performance. The study shows that the ML method can effectively optimize HT processes, leading to enhanced creep resistance and performance. The results indicate that the inclusion of physical features is crucial for improving the accuracy and interpretability of ML models in predicting creep rupture life. The study provides novel insights for enhancing the precision of predicting creep rupture life in superalloys and offers auxiliary support for the application of ML in the design of HT processes. The approach is validated through experimental data, demonstrating its effectiveness in optimizing HT processes and improving creep performance. The study also discusses the role of new features in classification models, the limitations of conventional SVR models, and the effectiveness of key feature screening in improving model accuracy. The results highlight the importance of HT parameters such as StT, AT, and PdT in determining creep rupture life and provide a framework for optimizing HT processes in superalloys. The study concludes that the proposed ML approach offers an efficient and effective means to evaluate the influence of HT processes on the creep properties of superalloys while significantly improving the accuracy of creep rupture life prediction.