How to Do Full-Text Reviews and Evidence Synthesis (With AI)

Описание: Full-text reviews are where literature reviews get super slow. Large systematic reviews can take 16+ months - and full-text review is the biggest chunk of that time.

In this session, I break down what full-text reviews actually involve (screening, data extraction, risk of bias), where AI can meaningfully reduce workload, and how evidence synthesis fits into the process. I also cover limitations and best practices for keeping humans in the loop.

This is part 3 of a 4-part series: AI Across the Research Stack.

TIMESTAMPS:
0:00 The Research Funnel (Recap)
1:55 Screening vs Full-Text Review
3:05 Three Components of Full-Text Review
4:28 What Full-Text Screening Looks Like (Rayyan)
5:16 Data Extraction Example (Covidence)
6:11 Risk of Bias Assessments
7:13 Main Issues with Manual Full-Text Review
8:28 Best AI Use Cases for Full-Text
10:16 Annotations in moara.io
12:11 Time Savings Potential (99.6%?)
13:13 What Is Evidence Synthesis?
15:02 Evidence Synthesis Output Example
16:19 AI for Qualitative vs Quantitative Synthesis
19:14 Evidence Synthesis in moara.io
19:55 Limitations and Best Practices
22:05 Wrap-Up

Key takeaway: AI is best at automating data extraction and cataloging — not replacing the analysis. Keep humans in the loop.

Part 1 (Search Strategy):    • Why Literature Search Strategies Should Be...  
Part 2 (Screening):    • Why Most Literature Reviews Skip Screening...  

Try moara.io: https://moara.io