The rapid digitization of electronic medical record (EMR) text notes has notably improved data analysis and patient care but also introduced data privacy challenges. Named Entity Recognition (NER) plays a vital role in protecting sensitive health information (SHI) during healthcare data analytics. This paper introduces a rule-based NER system designed to identify SHI entities within EMR text notes. Compared to machine learning-based (ML) systems, our approach efficiently adapts to the dynamic medical language and new terminologies in EMR text notes. The system primarily focuses on extracting SHI entities and normalizing temporal entities. We developed a rule-based NER system using a dataset of 1,734 training and 560 testing EMR text notes. The system employs HashMap and HashSet to categorize 18 SHI entities. The system features an information extraction algorithm that integrates contextual and linguistic rules analysis for precise and adaptable recognition. Our approach includes iterative optimization for refining the entity-specific patterns and improving contextual rules. Performance is evaluated through precision, recall, and F1-score metrics. Our results demonstrate an iterative optimization process of updating linguistic rules contributes to the system’s enhanced performance, increasing the F1 score from 0.63 to 0.92. In SHI entity recognition, the system achieved an average F1 score of 0.93. For temporal entity normalization, the system maintained high performance with an average F1-score of 0.90. In an ML-dominated NER field, our rule-based system not only demonstrates the adaptability and effectiveness of rule-based approaches but also provides a framework that effectively integrates new linguistic rules. Additionally, it highlights the potential of hybrid models, combining rule-based and ML methods for future research. This could lead to improved pattern recognition and increased efficiency in EMR text notes.