Background: Patient satisfaction is a crucial indicator of healthcare quality, which encompasses multiple dimensions of patient care. This research analyzes survey data from 720,000 patients across 42 Integrated Care Systems in the UK on patients' experiences with their GP practices by applying six machine learning regression algorithms to assess key drivers of patient satisfaction. Methods: We used Random Forest, XGBoost, CatBoost, Lasso, Ridge, and Elastic Net regression models. Cross-validation and hyperparameter tuning were conducted to optimize model performance and prevent overfitting. Results: Receptionist help emerged as the most significant factor influencing patient satisfaction, followed by listening skills of healthcare professionals and efficient appointment scheduling. Ridge and Elastic Net regression models demonstrated better performance, with R-squared values of 0.964, 0.956, respectively. Conclusions: Focusing on receptionist training, improving communication skills of healthcare professionals, and optimizing appointment scheduling systems can significantly increase patient satisfaction. Machine learning provides valuable information to improve healthcare quality.