Simplifying Cochrane Systematic Reviews with AI for Improved Shared Decision Making

ID: 

P - 53

Date: 

Monday 4 September 2023 - 12:30 to 14:00

Location: 

All authors in correct order:

Jahanfar S1, Hanumantha S2
1 Director of Cochrane Affiliate at Tufts School of Medicine, USA
2 Fellow Cochrane Affiliate at Tufts School of Medicine, USA
Presenting author and contact person

Presenting authors:

Shayesteh Jahanfar

Contact person:

Abstract text
Background: Cochrane systematic reviews provide high-quality, evidence-based information for healthcare decision-making. However, the language used in these reviews can be complex, making it difficult for patients and nonexperts to understand. This can hinder shared decision-making, where patients are equipped with the necessary information to make informed decisions about their health and treatment options. To facilitate shared decision-making and better communication with patients, it is important to simplify the information contained in Cochrane systematic reviews.
Objectives: This study aimed to evaluate the use of artificial intelligence (AI) in simplifying the lay language summaries of Cochrane systematic reviews to an 8th-grade level. The study compared the readability and accuracy of the simplified summaries generated by AI with those created by human experts and with the original Cochrane systematic reviews.
Methods: A sample of Cochrane systematic reviews was used in the study. Lay language summaries were created both by human experts and through an AI-powered simplification process to an 8th-grade level. The readability and accuracy of the summaries were assessed using established metrics.
Results: The results showed that the 8th-grade-level lay language summaries generated by AI were more accessible and understandable for patients and the general public without sacrificing the accuracy and comprehensiveness of the information. The simplified summaries had higher readability scores and were more accurately understood by patients and nonexperts compared with the original Cochrane systematic reviews.
Conclusions: The study highlights the importance of simplifying evidence-based information for shared decision-making and better communication between healthcare providers and patients. Using AI to generate 8th-grade-level lay language summaries of Cochrane systematic reviews is an effective way to achieve this goal. By making the information more accessible, patients can make informed decisions about their health and treatment options, leading to improved health outcomes. The study demonstrates the potential for AI to support shared decision-making by simplifying complex scientific information into understandable language for patients and the general public.