Gastroenterology Research, ISSN 1918-2805 print, 1918-2813 online, Open Access
Article copyright, the authors; Journal compilation copyright, Gastroenterol Res and Elmer Press Inc
Journal website https://gr.elmerpub.com

Original Article

Volume 18, Number 5, October 2025, pages 247-253

Enabling Drug-Induced Liver Injury Surveillance Through Automated Medication Extraction From Clinical Notes: A Medical Information Mart for Intensive Care IV Real-World Large Language Models Validation Study

Figures

Figure 1.
Figure 1. System architect. MIMIC-IV: Medical Information Mart for Intensive Care IV; DILI: drug-induced liver injury; LLM: large language models.
Figure 2.
Figure 2. Study flowchart. MIMIC-IV: Medical Information Mart for Intensive Care IV; DILI: drug-induced liver injury; LLM: large language models.

Tables

Table 1. Descriptive Statistics for the Data Cohort
 
Data sourceDescriptionTypeSample size usedTotal medications in sampleSource type
MIMIC-IV: Medical Information Mart for Intensive Care IV; DILI: drug-induced liver injury; N/A: not applicable.
MIMIC-IVSampled discharge summaries of MIMIC-IVUnstructured data1001,236Deidentified data
Combined DILI databaseMerged DILIrank/LiverTox dataReference databaseN/AAbout 1,200 unique entries (estimated)Reference databases

 

Table 2. Performance Metrics for Dataset
 
MIMIC-IV dataset
MIMIC-IV: Medical Information Mart for Intensive Care IV.
Total discharge summary100
Total medications1,236
True positive996
False positive174
False negative0
Precision0.85
Recall1.00
F1-score0.92