Back

Efficient Citation Screening by Weak Classifier Ensemble

2026-01-08 health informatics Title + abstract only
View on medRxiv
Show abstract

Citation screening in systematic review is time-consuming. Machine learning can help semi-automate it but faces obstacles. Each systematic review is a new dataset without initial annotations. Extreme class imbalance against irrelevant studies makes it difficult to select a good subset of samples to train a classifier. The rigid requirement of a (near) total recall of relevant studies demands a careful trade-off between accuracy and recall. This paper pilots a weak classifier ensemble approach to...

Predicted journal destinations

1
Journal of the American Medical Informatics Association
53 training papers
Top 0.3% 17.9%
2
Journal of Biomedical Informatics
37 training papers
Top 0.5% 9.1%
3
Journal of Medical Internet Research
81 training papers
Top 0.8% 9.1%
4
PLOS Digital Health
88 training papers
Top 2% 7.9%
5
JAMIA Open
35 training papers
Top 1% 6.5%
6
Scientific Reports
701 training papers
Top 44% 6.0%
7
BMC Medical Informatics and Decision Making
36 training papers
Top 2% 6.0%
8
PLOS ONE
1737 training papers
Top 81% 5.1%
9
International Journal of Medical Informatics
25 training papers
Top 0.8% 3.9%
10
BMC Medical Research Methodology
41 training papers
Top 0.5% 3.9%
11
npj Digital Medicine
85 training papers
Top 8% 3.1%
12
Research Synthesis Methods
17 training papers
Top 0.3% 2.0%
13
JMIR Medical Informatics
16 training papers
Top 2% 2.0%
14
Computers in Biology and Medicine
39 training papers
Top 3% 2.0%
15
Patterns
15 training papers
Top 0.2% 1.6%
16
Scientific Data
30 training papers
Top 2% 1.2%
17
Biology Methods and Protocols
19 training papers
Top 0.8% 1.0%
18
Bioinformatics
24 training papers
Top 3% 0.7%
19
Frontiers in Digital Health
18 training papers
Top 4% 0.7%
20
JMIR Formative Research
31 training papers
Top 7% 0.7%
21
Journal of Clinical Epidemiology
29 training papers
Top 3% 0.7%
22
BMJ Open
553 training papers
Top 67% 0.7%
23
Nature Communications
483 training papers
Top 65% 0.5%