SEPLN PhD Symposium 2024 – Automatic Medical Knowledge Graph Construction

Welcome to the repository for the PhD Symposium 2024 submission by Joel Pardo-Ferrera, titled "Automatic Medical Knowledge Graph Construction". This repository contains the essential materials presented at the symposium.

Repository Structure and Files

• Paper: SEPLN_PHDSympos_2024_Joel_pardo_ferrera2.pdf

  • This is the full paper submitted for the symposium, outlining the methodology, objectives, and key findings of the research.

• Presentation: SEPLN_Sympos_definitiva.pdf

  • A slide deck summarizing the key points of the research, designed for a 3-minute oral presentation at the symposium.

• Poster: PhD_SEPLNSympos2024_jpardo_definitivo.pdf

  • A poster highlighting the main contributions and findings of the research, used for the symposium's poster session.

Abstract

Knowledge Graphs are crucial for structuring and integrating large amounts of data, improving decision- making and data interoperability, especially in the healthcare domain. This PhD thesis aims to implement a unified end-to-end framework for building a cross-lingual KG for English and Spanish in the healthcare sector using NLP techniques. Addressing the reliance on traditional methods in KG construction and the limited non-English language resources, this work seeks to refine the information extraction process within unstructured medical texts and facilitate the (re)use of existent ontologies (schema to represent the real-world).

Research Focus

Developing an automated framework to generate domain-specific KGs for the electronic health sector, aimed at assisting medical professionals in the decision-making process.

1. Preliminary Health Domain Analysis and Planning:

• Initial analysis by gathering requirements from healthcare professionals and analyzing Electronic Health Records (EHRs) authored by them.
• This stage is critical to understanding the data's structure and developing methodologies for extracting knowledge from unstructured clinical text.

2. Ontology Review, Selection, and Adaptation:

• Review and select existing ontologies within the health domain, aligning them with the requirements identified in the preliminary analysis.
• The most suitable ontology is chosen, and any gaps in addressing specific health domain needs are identified.
• The selected ontology is adapted (if it is needed) to model the entire problem comprehensively.
• This step ensures that the constructed Knowledge Graph (KG) will be interoperable and contextually accurate for the medical domain.

3. Technological Framework Development:

• Fine-tuning Large Language Models (LLMs) to incorporate the ontology knowledge, enabling the model to identify relevant entities and relationships within the clinical text.
• Structure the extracted data into RDF triplets, forming a formal and usable Knowledge Graph (KG).

4. Evaluation Framework Development:

• Evaluation framework to assess the entire workflow and the performance of individual IE tasks.
• A test environment is created where healthcare professionals can input real-world cases, allowing for the automatic extraction and validation of KGs in decision-making scenarios.
• This ensures the practical applicability and accuracy of the KG in assisting healthcare professionals.

Funding

This work has been funded by INESData (Infraestructura para la INvestigación de ESpacios de DAtos distribuidos en UPM) project, from Ministerio para la Transformación Digital y de la Función Pública (PRTR, UNICO I+D CLOUD, EU NextGeneration).

Installation

If you'd like to explore the paper and materials, simply clone the repository:

git clone https://github.com/JPardo08/SEPLN2024.git