Back

The Effect of LLM Assistance on Diagnostic Accuracy: A Meta-Analysis

2025-12-12 health informatics Title + abstract only
View on medRxiv
Show abstract

Large language models (LLMs) are increasingly used in clinical settings, yet their effect on diagnostic accuracy of physicians has not been systematically quantified. We conducted a systematic review and meta-analysis of studies analyzing LLM-assisted diagnosis published between January 2020 and June 2025. Across 15 studies (43 effect sizes; 498 physicians; 7,274 case evaluations), LLM assistance significantly improved diagnostic accuracy compared to physicians without LLM support (Hedges g = 0....

Predicted journal destinations

1
Journal of the American Medical Informatics Association
53 training papers
Top 1% 14.2%
2
BMC Medical Informatics and Decision Making
36 training papers
Top 0.2% 8.8%
3
PLOS Digital Health
88 training papers
Top 2% 7.6%
4
npj Digital Medicine
85 training papers
Top 2% 7.6%
5
JAMIA Open
35 training papers
Top 2% 6.2%
6
PLOS ONE
1737 training papers
Top 78% 5.8%
7
Journal of Medical Internet Research
81 training papers
Top 3% 5.8%
8
Scientific Reports
701 training papers
Top 53% 5.4%
9
Journal of Biomedical Informatics
37 training papers
Top 3% 4.1%
10
BMJ Open
553 training papers
Top 38% 4.1%
11
International Journal of Medical Informatics
25 training papers
Top 1% 3.8%
12
JMIR Medical Informatics
16 training papers
Top 0.7% 3.1%
13
BMC Medical Research Methodology
41 training papers
Top 2% 1.9%
14
JAMA Network Open
125 training papers
Top 10% 1.9%
15
Computers in Biology and Medicine
39 training papers
Top 5% 1.5%
16
JMIR Formative Research
31 training papers
Top 2% 1.5%
17
Frontiers in Digital Health
18 training papers
Top 2% 1.0%
18
JMIR Public Health and Surveillance
45 training papers
Top 11% 0.7%
19
Nature Communications
483 training papers
Top 61% 0.7%
20
Frontiers in Public Health
135 training papers
Top 34% 0.7%
21
The Lancet Digital Health
25 training papers
Top 2% 0.7%
22
Journal of Clinical Epidemiology
29 training papers
Top 3% 0.7%
23
Patterns
15 training papers
Top 4% 0.7%