This study presents a comprehensive medical natural language processing pipeline that combines advanced entity extraction, syntactic relation mining, domain-specific embeddings and temporal parsing to transform annotated biomedical abstracts into structured clinical information. The pipeline framework uses a modular workflow that includes data parsing, spaCy-based preprocessing, exploratory corpus analysis, BioBERT-driven NER, dependencybased relation extraction and rule-based temporal detection to handle 593 annotated abstracts with inline XML-style biological entities. A BioBERTextracted collection of 3,490 entity mentions from the first 100 documents is produced by the pipeline, which identifies 4,424 annotated entities with diseaserelated categories predominating the corpus structure. Mapping disease-genetreatment relationships with a limited number of temporal indications is efficient, as demonstrated by the 613 syntactic relationships generated by relation extraction and the 28 duration expressions captured by temporal parsing. According to co-occurrence analysis, there is a substantial correlation between modifier and specific disease, which reflects typical clinical narrative patterns. This study presents a biomedical NLP pipeline for extracting entities, finding relations and parsing time. It aims to transform annotated biomedical text into structured clinical knowledge. However, it also points out limitations in time coverage and adaptability