Automation of Article Selection Process in Systematic Reviews Through Artificial Neural Network Modeling and Machine...

Area of Science:

• Health Sciences
• Computer Science
• Information Science

Background:

• Systematic reviews require extensive literature searches, with article selection being a time-consuming bottleneck.
• Machine learning and artificial intelligence offer potential solutions for automating repetitive research tasks.
• Efficient article selection is crucial for the timely completion of systematic reviews.

Purpose of the Study:

• To develop artificial neural network (ANN) models for automating article selection in systematic reviews.
• To focus the application on systematic reviews concerning "Mindfulness and Health Promotion."
• To create a tool that aids researchers by automating the article selection process.

Main Methods:

• The study utilizes Python programming for system development.
• Key steps include data import, duplicate and non-article exclusion, ANN model creation, model comparison, and system sharing.
• The system's effectiveness will be tested using systematic reviews from "Mindfulness and Health Promotion" and "Orthopedics."

Main Results:

• The project is scheduled for completion in December 2021, with results expected by March 2022.
• The development will result in an automated article selection system.
• The system will be evaluated for its effectiveness in identifying relevant articles.

Conclusions:

• An automated system with adjustable sensitivity will be created for selecting scientific articles in systematic reviews.
• The system is designed to be adaptable for application across various scientific fields.
• Results and models will be shared via the "Observatory of Evidence" in public health to support researchers.