With medical digitalization, online patient reviews have become crucial for evaluating healthcare quality. Traditional surveys face limitations in sample size and response rate, hindering real-time patient satisfaction assessment. This study develops an AI-driven Healthcare Decision Support System (AI-DSS) using NLP and ML to analyze patient reviews from NTUH and TVGH, identifying key satisfaction factors. Using BERT for sentiment analysis and text mining, feedback is categorized into five dimensions: facility environment, waiting time, staff attitude, medical procedures, and treatment outcomes. Multiple linear regression quantifies their impact on satisfaction, while a ChiSquare Test compares hospital performance. Results show waiting time and staff attitude as critical factors, with significant differences in the facility environment and staff attitude (p < 0.001). This study confirms AI's role in real-time satisfaction monitoring and data-driven healthcare improvements
