Related Experiment Video
Updated: Sep 4, 2025
10:33
Expedited Radiation Biodosimetry by Automated Dicentric Chromosome Identification ADCI and Dose Estimation
Published on: September 4, 2017
15.9K
DxGenerator: An Improved Differential Diagnosis Generator for Primary Care Based on MetaMap and Semantic Reasoning
Ali Sanaeifar1, Saeid Eslami1, Mitra Ahadi2
1Department of Medical Informatics, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran.
Methods of Information in Medicine
|July 20, 2022
Summary
A new web-based differential diagnosis tool, DxGenerator, improved diagnostic accuracy for primary care physicians. It significantly enhanced the ranking of correct diagnoses, especially for uncommon gastrointestinal diseases.
Area of Science:
- Medical Informatics
- Clinical Decision Support Systems
- Artificial Intelligence in Healthcare
Background:
- Medical errors are a significant cause of death, prompting the development of computerized interventions.
- Differential diagnosis generators aid primary care physicians by suggesting potential diagnoses from initial symptoms.
- The accuracy of these systems is often measured by the rank of the correct diagnosis in the generated list.
Purpose of the Study:
- To design and evaluate a novel, practical web-based differential diagnosis generator for primary care.
- To enhance diagnostic accuracy in primary care settings through an improved decision support system.
Main Methods:
- Developed DxGenerator, an online clinical decision support system integrating a semantic database with the Unified Medical Language System (UMLS) knowledge base via MetaMap and natural language processing.
- Modeled 120 gastrointestinal diseases causing abdominal pain into the database.
- Evaluated DxGenerator using 172 patient vignettes, comparing its performance against ISABEL, a leading similar system, using the Wilcoxon signed-rank test.
Main Results:
- DxGenerator improved the mean position of correct diagnoses from 4.2 ± 5.3 in ISABEL to 3.2 ± 3.9 across 172 vignettes.
- This improvement in diagnostic accuracy was statistically significant (p < 0.05) in the subgroup of uncommon diseases.
Conclusions:
- Leveraging the UMLS knowledge base and MetaMap tools enhances the accuracy of diagnostic systems that accept free-text input.
- The proposed methods can improve the adoption and acceptance of medical diagnostic systems within the healthcare community.
