Related Experiment Video
Updated: Jul 19, 2026
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.5K
Assessing the article screening efficiency of artificial intelligence for Systematic Reviews
Yu-Ting Chan1, Jilaine Elliscent Abad1, Serge Dibart1
1Department of Periodontology, Henry M. Goldman School of Dental Medicine, Boston University, 635 Albany Street, Boston, MA 02118, United States.
Journal of Dentistry
|July 27, 2024
Summary
Artificial intelligence (AI) tools like ASReview significantly improve systematic review efficiency by prioritizing articles. This AI program can save approximately 60% of the time and effort typically needed for article screening.
Area of Science:
- Medical informatics
- Artificial intelligence in healthcare
- Systematic review methodology
Background:
- Machine learning (ML) and AI tools enhance efficiency in medicine and academia.
- ASReview is an AI program designed to streamline systematic reviews by automating article prioritization.
- This study evaluates ASReview's screening efficiency and influencing factors.
Purpose of the Study:
- To examine the screening efficiency of ASReview in systematic reviews.
- To identify factors influencing ASReview's efficiency.
Main Methods:
- Searched six periodontics topics in PubMed and Web of Science.
- Trained ASReview with relevant/irrelevant articles for ML optimization.
- Evaluated screening efficiency based on normalized non-reviewed articles and time expenditure.
Main Results:
- An average of 60.2% of articles did not require extensive screening.
- All relevant articles were identified within the first 39.8% of reviewed publications.
- No significant efficiency variations were found with different training methods or article ratios.
Conclusions:
- ASReview provides an average 60.2% improvement in screening efficiency due to ML capabilities.
- Human discernment remains crucial for training AI tools like ASReview effectively.
- ASReview has the potential to save approximately 60% of time and effort in article screening.
