Back

Large-Language Models for data extraction from written kidney biopsy reports

2026-02-25 pathology Title + abstract only
View on medRxiv
Show abstract

IntroductionKidney biopsy reports contain rich information that is clinically actionable and useful for research. However, the narrative format hinders scalable reuse. We here investigated whether open-source large language models (LLMs) can extract relevant, standardized readouts from native kidney biopsy pathology reports. MethodsGerman free-text native kidney biopsy reports were parsed with three open-source LLMs (Llama3 70B, Llama3 8B, MedGemma) to generate structured JSON outputs covering ...

Predicted journal destinations

1
Scientific Reports
701 training papers
Top 13% 11.7%
2
PLOS ONE
1737 training papers
Top 53% 11.7%
3
npj Digital Medicine
85 training papers
Top 3% 6.4%
4
Nature Communications
483 training papers
Top 16% 6.4%
5
BMJ Open
553 training papers
Top 37% 4.1%
6
JAMA Network Open
125 training papers
Top 4% 3.8%
7
Journal of the American Society of Nephrology
19 training papers
Top 0.9% 3.1%
8
Cureus
64 training papers
Top 4% 2.8%
9
International Journal of Medical Informatics
25 training papers
Top 3% 1.9%
10
Cancers
57 training papers
Top 5% 1.9%
11
PLOS Digital Health
88 training papers
Top 10% 1.9%
12
Journal of Clinical Microbiology
77 training papers
Top 3% 1.9%
13
The Lancet Digital Health
25 training papers
Top 0.2% 1.9%
14
Computers in Biology and Medicine
39 training papers
Top 5% 1.5%
15
Med
26 training papers
Top 0.2% 1.3%
16
Diabetologia
23 training papers
Top 2% 1.3%
17
eLife
262 training papers
Top 45% 1.0%
18
BMC Medicine
155 training papers
Top 28% 0.9%
19
BMC Cancer
21 training papers
Top 5% 0.9%
20
JCI Insight
63 training papers
Top 7% 0.9%
21
European Respiratory Journal
44 training papers
Top 5% 0.7%
22
PLOS Neglected Tropical Diseases
166 training papers
Top 21% 0.7%
23
Frontiers in Medicine
99 training papers
Top 24% 0.7%
24
iScience
74 training papers
Top 13% 0.7%
25
The Journal of Molecular Diagnostics
24 training papers
Top 3% 0.7%