Named Entity Recognition (NER) is an important element of healthcare Natural Language Processing (NLP), it aids in extracting the necessary structured clinical entities such as diseases, drugs, symptoms, and procedures from unstructured Electronic Health Records (EHRs) and biomedical texts. This survey explores recent developments in healthcare-specific NER, classifying methods as traditional, deep learning-based, and large language model (LLM) driven approaches. This paper refers to recent work (2022–2025) and a comparative assessment based on methodologies, datasets, known entity types, and evaluation metrics. Results show substantial performance gain with transformer-based and few-shot learning methods, especially when used in conjunction with domain-specific pre-trained models and prompt engineering. The paper also identifies upcoming challenges and directions in healthcare NER.