A web-based tool for automatically linking clinical trials to their publications

Area of Science:

• Biomedical Informatics
• Clinical Trial Management
• Publication Tracking

Background:

• Evaluating clinical trial outcomes requires linking registry data with publications.
• Manual searching for associated publications is time-consuming and unreliable.

Purpose of the Study:

• To develop a statistical model for automatically identifying PubMed articles reporting clinical outcome results from registered trials.
• To create a publicly accessible tool to aid researchers in this process.

Main Methods:

• A machine learning model was trained using pairs of registered trials and their known publications.
• Features included matching metadata between PubMed articles and trial registry fields, and known publication links.

Main Results:

• The model effectively ranked known linked articles, with a median best rank of 1.0.
• 91% of known linked publications were ranked within the top 10 results.

Conclusions:

• The developed model and web tool automate the identification of clinical trial publications.
• This facilitates the study of trial outcomes and their relationship to original trial designs.