Skip to main content

Converting Medical Guidelines into Knowledge Graph

Primary supervisor

Ehsan Shareghi


Medical guidelines provide human experts with steps on conducting diagnosis. While existing online knowledge graphs such as Unified Medical Language System (UMLS) [1] provide a wide coverage of various biomedical entities, their coverage is limited in certain specific domains such as Ophthalmology. In this project we will build an automated system that convert the offline medical guidelines of Glaucoma [2] and Diabetic Retinopathy [3] to machine readable knowledge graphs. We will follow the FHIR schema [4] to store the constructed knowledge graph for these two disorders and link them (when possible) to existing entities from UMLS. The student will learn, analyze, and develop models for two fundamental and critical NLP problems in this project: medical Entity Recognition and and Relation Extraction (e.g., SapBERT [5]) and will have the opportunity to share the findings with world-renowned Ophthalmologist (Professor Hiroshi Ishikawa), and work closely with Dr Bhavna Antony (former scientist with IBM, current research coordinator with Alfred Hospital). The findings of the work will be submitted as publication to Bioinformatics.



Student cohort

Single Semester
Double Semester


[1] The Unified Medical Language System (UMLS): integrating biomedical terminology, Olivier Bodenreider, Nucleic Acids Research, Volume 32, January 2004




[5] Self-alignment Pre-training for Biomedical Entity Representations


Required knowledge

  • Proficiency in Python is required
  • Working knowledge of NLP (e.g.,  Named Entity Recognition and Relation Extraction) is required
  • Familiarity with the gensim and pytorch libraries is desired
  • Familiarity with Text-based Transformer models and HuggingFace is desired
  • Very good verbal and written communication skill is required