Clinical NLP

Clinical NLP studies how to process medical records, radiology reports, research articles, and patient-clinician conversations for tasks such as entity recognition, coding, screening, and summarization. Common approaches include domain-adapted BERT encoders, fine-tuned or prompted large language models, retrieval and knowledge-graph methods, and rule-based or hybrid systems. Work increasingly covers multilingual data, synthetic corpora, privacy-preserving processing, and human-AI workflows, with evaluation spanning clinical benchmarks and real-world settings. A central tension is that strong scores on one dataset or language may not transfer across hospitals or patient groups, while fluent or accurate-looking outputs still require checks for factuality, fairness, and clinical safety.