Systematic Reviews, Scoping Reviews, and Health Technology Assessments - Searching the Literature

Generative AI tools cannot replace the human in any tasks required for evidence synthesis projects. Humans must be kept in the loop to validate all AI outputs. AI tools may be useful as starting points but human expertise is required to review and contextualize outputs.  

Below are several resources that outline the potential of using AI in reviews more generally. In the sections that follow, you can view:

• AI tools for Question Formulation and Methodology Selection
• Searching the Literature

AI and Evidence Synthesis (General Resources)

For Question Development and Methodology Selection:

AI text-generator (Generative AI) tools including ChatGPT, Claude, Copilot, Google Gemini, and Perplexity can be used to assist with these initial stages of evidence synthesis. 

Question Formulation
The research question is the foundation of any research project and it is vital for research teams to formulate a well-defined question to clearly outline the aims of the project. Generative AI tools can be used to help with tasks such as: 

• Developing PICO framework
• Drafting inclusion and exclusion criteria
• Exploring new areas, getting familiar with your topic

Sample Prompts:

• “I’m interested in starting a research project related to exercise and smoking cessation. What are some important things to consider?”
• “Here is my PICOS question framework. Please provide me with feedback on each element, particularly on concepts that could be made clearer or more specific” 

Select Review Methodology 
There are many different types of reviews to choose from and they include critical reviews, narrative reviews, scoping reviews, systematic reviews, and umbrella reviews to name a few. Systematic reviews may not be the best suited review type for all research questions. Therefore, selecting the appropriate review methodology is one of the important decisions that must be made in the early stages of the project.

Sample Prompts: 

• “I’m interested in exploring x. Which review methodology should I use if I were to embark on a research project?”
• “I would like to conduct a meta-analysis but there is a lot of heterogeneity in the study estimates and in the study designs. How much is too much heterogeneity and how can I evaluate it qualitatively and quantitatively? Can you list some references where I can find more information to guide my synthesis?” 
   

For the Literature Search: 

Due to the labour intensive search development process that librarians and other information professionals may experience, there may be cost/time saving measures by integrating AI tools into the workflow.

• Generate search terms
• Assist with Boolean logic
• “Peer” review search strategies
• Draft search documentation - create templates for documenting search strategies including databases searched, dates, and search strings
• Refine search strategies
• Search translation
• Identify grey literature sources

HOWEVER! Tools such as ChatGPT, Perplexity, Consensus, etc. should be used to complement a traditional search process using appropriate databases. No AI tool yet replaces a human-powered search, especially when as it relates to transparency and reproducibility. 

Below is a link to a public Zotero library. You can browse articles in the library or download the citations to your own reference manager to read later.