ZebraMap: A Multimodal Rare Disease Knowledge Map with Automated Data Aggregation & LLM-Enriched Information Extraction Pipeline

Sep 30, 2026

Uploaded by Riley

Rare diseases frequently face delayed diagnosis because clinical evidence is scattered across unstructured literature, individual case reports, and heterogeneous data formats. ZebraMap is a multimodal knowledge map designed to consolidate rare disease information and convert narrative case evidence into structured, machine-readable data. It uses Orphanet to identify 1727 rare diseases, links them to PubMed case reports, and applies an automated LLM structuring pipeline plus retrieval-augmented generation to enrich missing disease-level descriptions. The resulting dataset includes 69,146 structured patient-level case texts and 98,038 linked clinical images, supporting computable research.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages26

Reading Time1 h 6 min

AccessFree to download

Abstract

Introduction

Background

Article ZebraMap: A Multimodal Rare Disease Knowledge Map with Automated Data Aggregation & LLM-Enriched Information Extraction Pipeline Md. Sanzidul Islam 1,2, *, Amani Jamal 1,3 and Ali Alkhathlan 1 Academic Editor: Ayman El-Baz Received: 18 November 2025 Revised: 4 December 2025 Accepted: 13 December 2025 Published: 29 December 2025 Copyright: © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license. 1 Department of Computer Science, FCIT, King Abdulaziz University, Jeddah 21589, Saudi Arabia; [atjamal@kau.edu.sa](atjamal@kau.edu.sa) (A.J.); [analkhathlan@kau.edu.sa](analkhathlan@kau.edu.sa) (A.A.) 2 Department of Computer Science, FSIT, Daffodil International University, Birulia 1216, Bangladesh 3 Center of Research Excellence in Artificial Intelligence and Data Science, King Abdulaziz University, Jeddah 21589, Saudi Arabia * [Correspondence: mislam0008@stu.kau.edu.sa or](Correspondence: mislam0008@stu.kau.edu.sa or) [sanzid.cse@diu.edu.bd](sanzid.cse@diu.edu.bd); Tel.: +966-558393507 Abstract Background: Rare diseases often lead to delayed diagnosis because clinical knowledge is fragmented across unstructured research, individual case reports, and heterogeneous data formats. This study presents ZebraMap, a multimodal knowledge map created to consolidate rare disease information and transform narrative case evidence into structured, machine-readable data. Methods: Using Orphanet as the disease registry, we identified 1727 rare diseases and linked them to PubMed case reports. We retrieved 36,131 full-text case report articles that met predefined inclusion criteria and extracted publication metadata, patient demographics, clinical narratives (cases), and associated images. A central methodological contribution is an automated large language model (LLM) structuring pipeline, in which free-text case reports are parsed into standardized fields, such as symptoms, diagnostic methods, differential diagnoses, treatments, and outcome that produce structured case representations and image metadata matching the schema demonstrated in our extended dataset. In parallel, a retrieval-augmented generation (RAG) component generates concise summaries of epidemiology, etiology, clinical symptoms, and diagnostic techniques by retrieving peer-reviewed research to enhance missing disease-level descriptions. Results: The final dataset contains 69,146 structured patient-level case texts and 98,038 clinical images, each linked to a particular patient ID, disease entry, and publication. Overall cosine similarity between curated and generated text is 94.5% and performance in information extraction and structured data generation is satisfactory. Conclusions: ZebraMap provides the largest openly accessible multimodal resource for rare diseases and enables data-driven research by converting narrative evidence into computable knowledge. Keywords: multimodal dataset; rare disease diagnosis; knowledge map; case reports; LLM applications; retrieval-augmented generation (RAG); clinical images; structured clinical records 1. Introduction Rare diseases, as defined in the European Union, are those affecting fewer than 1 in 2000 individuals [1]; and rare diseases, though individually uncommon, collectively affect an estimated 3.5–5.9% of the global population, amounting to 263–446 million people worldwide at any given time [2,3] . According to a comprehensive analysis of the Orphanet database, there are over 6000 distinct rare diseases, with approximately 72% being of genetic origin and 70% manifesting in childhood [1] . These conditions are individually infrequent but, in aggregate, present a substantial global health burden, contributing to significant morbidity, mortality, and unmet medical needs. Misdiagnosis and diagnostic delays are common challenges for people with rare diseases. Studies

FAQs

What problem does ZebraMap address in rare disease research?

ZebraMap targets delayed diagnosis caused by fragmented clinical knowledge across unstructured reports and inconsistent data formats, making it hard to derive computable evidence.

How does ZebraMap build its dataset?

It uses Orphanet to identify 1727 rare diseases, links them to PubMed case reports, retrieves qualifying full-text articles and extracts publication metadata, patient demographics, clinical narratives, and associated images.

What are the main AI components used for information extraction?

ZebraMap includes an automated LLM structuring pipeline to parse free-text into standardized fields, and a retrieval-augmented generation (RAG) component to generate concise summaries by retrieving peer-reviewed research.

Sep 30, 2026

Uploaded by Riley