1 of 55

Literature Reviews: Resources, Best Practices, and AI Tools

May 13, 2026

MHPE Webinar

Emily Capellari, Associate Director and Informationist, Taubman Health Sciences Library

2 of 55

Learning objectives

After today's webinar, attendees will be able to:

Describe key guidelines and resources that support the literature review process.
Apply literature searching techniques and best practices to effectively conduct literature reviews.
Discuss considerations and caveats for using artificial intelligence (AI) tools in literature reviews.

2

3 of 55

Literature Reviews Overview

"Literature review" is used in so many ways!

As a section in primary research articles
As a title of narrative reviews
As a general term for advanced evidence synthesis methodologies:

Scoping reviews
Systematic reviews (with and without meta-analysis)

The type of literature review you pursue should be suited to the project needs and your resourcing.

Consider the rigor and transparency needed, and time commitment involved.

3

4 of 55

A typology of reviews

Grant MJ, Booth A. A typology of reviews: an analysis of 14 review types and associated methodologies. Health Info Libr J. 2009 Jun;26(2):91-108.

4

5 of 55

A typology of reviews, 2 of 2

Grant MJ, Booth A. A typology of reviews: an analysis of 14 review types and associated methodologies. Health Info Libr J. 2009 Jun;26(2):91-108.

5

6 of 55

Key Guidelines

Methodology guidelines
Reporting guidelines
AI guidance

6

7 of 55

JBI Manual for Evidence Synthesis (2024 Edition)

7

Chapter 10: Scoping Reviews

8 of 55

Develop a protocol

8

9 of 55

PRISMA Reporting Guidelines

Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020
Key PRISMA extensions:

PRISMA for Protocols
PRISMA for Scoping Reviews

PRISMA guidelines encourage transparent, rigorous reporting

9

10 of 55

PRISMA Scoping Review Extension Checklist

PRISMA-Scoping Review Checklist

10

11 of 55

Using PRISMA Reporting Guidelines

PRISMA and its extensions are reporting guidelines
Use PRISMA in combination with methodology guidelines for your review type
Cite both guidelines in your manuscript

11

12 of 55

PRISMA Flow Diagram

Document:

Results from databases
Screening process
Included studies
Excluded records and studies

12

13 of 55

Responsible use of AI in Evidence Synthesis (RAISE 2026)

Recommendations and guidance on responsible use of AI in evidence synthesis
Three RAISE 2026: recommendations and guidance papers on Open Science Framework:

RAISE 1: recommendations for practice
RAISE 2: building and evaluating AI evidence synthesis tools
RAISE 3: selecting and using AI evidence synthesis tools

13

14 of 55

RAISE 3: selecting and using AI evidence synthesis tools

Excerpt from Abstract (page 1)

"...deciding when and how to use AI, and how to document its use transparently, is not straightforward. Technical, ethical and organisational challenges mean that evidence synthesists need to understand key risks and requirements before making informed decisions about adopting AI."

14

15 of 55

RAISE 3 - Recommendation Key

Table 1: Tool use recommendations (pp. 9-10)

Recommendation	Description
Acceptable for use	AI outputs may be used directly within the review workflow, if any limitations or potential biases are acknowledged and accounted for.
Human verification required	AI outputs may be used to support review tasks but must be carefully checked by humans before use. The degree of checking required may vary, but typically will require a human to read and possibly make amendments to the entirety of the output.
Requires validation within the review	AI outputs may be used if their performance is explicitly evaluated within the context of the review itself and deemed adequate (e.g., comparable to human performance).
Exploratory and supplementary use	AI outputs may be used for developing ideas or as a starting point to support understanding. All outputs should be extensively refined by human reviewers prior to use for a review task. Alternatively, outputs may be appropriate for use as an additional, supplementary approach, but without replacing established processes.
Not acceptable for use	The current state of technology means these AI outputs have such serious limitations, that they should not be relied upon.

16 of 55

RAISE 3 - Recs. by tool class

Table 2: current (February 2026) state of AI tools (pp. 11-17)

Organized by section and task
Example tools
Recommendation (from Table 1)

16

17 of 55

RAISE 3 - Recs. by tool class, cont.

Table 2: current (February 2026) state of AI tools (pp. 11-17)

Organized by section and task
Example tools
Recommendation (from Table 1)

17

18 of 55

RAISE 3 - Recs. by tool class, cont.

Table 2: current (February 2026) state of AI tools (pp. 11-17)

Organized by section and task
Example tools
Recommendation (from Table 1)

18

19 of 55

Other AI guidance

Degree or program requirements
Journal policies

Before using AI tools, carefully check the author guidelines of your target journals!

Transparent and accurate reporting of AI tools used

19

The Lancet's Information for Authors (PDF, p. 3)

20 of 55

Literature Searching Techniques and Resources

Search techniques
Search documentation
Library resources for literature reviews

20

21 of 55

Controlled vocabulary

Official subject headings for concepts
Database-specific

Medline - Medical Subject Headings
Embase - Emtree
CINAHL - CINAHL Headings
PsycInfo - APA Thesaurus of Psychological Index Terms

Because databases consider controlled vocabularies proprietary, AI search translation tools like Polyglot don't

21

"Learning Health System"[MeSH]

22 of 55

Controlled vocabulary

AI search translation tools like Polyglot don't translate controlled vocabulary well

Controlled vocabularies is database-specific and proprietary

Manually check and update controlled vocabulary terms if you use a translation tool!

22

Polyglot provides the MeSH official heading with Embase's syntax for Emtree terms

23 of 55

Controlled vocabulary hierarchy

Broader and narrower terms related to the Medical Subject Headings
PubMed explodes MeSH

"Health Services Accessibility"[MeSH] will search all narrower MeSH terms
If needed, use the checkbox to turn off auto exploding: "Health Services Accessibility"[Mesh:NoExp]

23

24 of 55

Search logic - OR between related terms

24

"Health equity"

OR

"Health disparities"

25 of 55

Search logic - AND between unique concepts

25

"Health equity"

AND

"Learning health system"

26 of 55

Boolean NOT for search testing

Avoid NOT in final search strategies for thorough scholarly projects due to the risk of removing relevant items.
Use NOT during search testing and development.

For example, look for additional relevant text words by searching for items indexed with relevant controlled vocabulary terms that do not use your known text words in the title and abstract.
Example in PubMed:�"Health Equity"[MeSH] NOT ("health equity"[tiab] OR "health disparity"[tiab] OR "health disparities"[tiab] OR "healthcare disparities"[tiab] OR "systemic bias"[tiab])

26

27 of 55

Example health equity search concept

("health equity" OR "health disparity" OR "health disparities" OR "health inequities" OR "healthcare disparities" OR "systemic bias" OR "equitable care" OR "equitable healthcare" OR "equitable health system" OR "equity-based health policy")

27

Related terms are combined with OR and are enclosed in parentheses to create a subset for the concept.

28 of 55

Advanced search technique: phrase searching

("health equity" OR "health disparity" OR "health disparities" OR "health inequities" OR "healthcare disparities" OR "systemic bias" … )

28

In most databases, use quotation marks " around multi-word search terms for exact phrase searching.

This makes your search more specific. In the example above, "systemic bias" removes an irrelevant results that mentions risk of bias in a study on patients with systemic sclerosis.

29 of 55

Add controlled vocabulary terms for the database

("Health Equity"[MeSH] OR "Healthcare Disparities"[Mesh] OR "Health Inequities"[Mesh] OR "health equity" OR "health disparity" OR "health disparities" OR "healthcare disparities" OR "health inequities" OR "systemic bias" OR "equitable care" OR "equitable healthcare" OR "equitable health system" OR "equity-based health policy")

29

30 of 55

Advanced search technique: field tags in PubMed

("Health Equity"[MeSH] OR "Healthcare Disparities"[Mesh] OR "Health Inequities"[Mesh] OR "health equity"[tiab] OR "health disparity"[tiab] OR "health disparities"[tiab] OR "healthcare disparities"[tiab] OR … )

30

The field tag [tiab] focuses the search on the title, abstract, and author supplied keyword fields.

In PubMed, the field tag must be added after each search term.

31 of 55

Advanced search technique: truncation

("Health Equity"[MeSH] OR "Healthcare Disparities"[Mesh] OR "Health Inequities"[Mesh] OR "health equity"[tiab] OR "health disparit*"[tiab] OR "healthcare disparit*"[tiab] OR "health inequit*"[tiab] OR … )

31

In most databases, the asterix symbol * searches for all endings of a root word, such as disparity and disparities.

In PubMed, the root word must be at least 4 characters long.

32 of 55

Advanced search technique: adjacency searching in PubMed

("Health Equity"[MeSH] OR "Healthcare Disparities"[Mesh] OR "Health Inequities"[Mesh] OR "health equity"[tiab:~2] OR "health disparity"[tiab:~2] OR "health disparities"[tiab:~2] OR … )

32

In PubMed, the adjacency operator is added to the [tiab] field tag following each exact phrase: "multi word phrase"[tiab:~n]. This allows n additional words to appear within the phrase.

In PubMed, adjacency and truncation CANNOT be combined.

33 of 55

Advanced search technique: adjacency searching in PubMed

https://pairwise-pubmed.streamlit.app/

33

34 of 55

Add unique concepts to the search with AND

("Health Equity"[MeSH] OR "Healthcare Disparities"[Mesh] OR "Health Inequities"[Mesh] …) AND ("Learning Health System"[Mesh] OR "learning health system*"[tiab] OR "learning healthcare system*"[tiab] OR "learning communit*"[tiab] OR ("continuous improvement"[tiab] AND (data[tiab] OR research[tiab] OR evidence[tiab] OR "electronic health record*"[tiab])))

34

35 of 55

35

Refine your search as needed and export the results.

36 of 55

Consider keeping notes about search testing you complete

36

Document the iterative search development process

37 of 55

Search Documentation

37

Even if you're working on a literature review that falls on the less formal, rigorous side, it can be helpful to keep track of your search strategies. You may want to return to them as your project progresses.

I like to create a Search Summary table that documents the numbers retrieved from each database. I'll add columns to the Search Summary for future search updates.

Below that, I use tables to keep track of each search concept block (in this PubMed example searches #1, #2, #3 and #4), combined search block (here row #5 which is #1 AND #2 AND #3 OR #4) and any limits I have applied. In this case, a publication date limit of 2009 to present. If you apply any limits, include your rationale for the limit in your search documentation. This should be reported in your Methods to help the audience understand the decisions that shaped your set of evidence screened for the review. In this case, there was legislation and corresponding healthcare practice changes for unhoused patients in 2009 so articles published before then would not be relevant to the scoping review.

When I include the Search Strategies in the supplemental materials for evidence synthesis publications, I remove the column of results numbers. I only report the Set #s and exact Search Strategies run in each database.

You may also prefer to use a numbered list format instead of tables -- use the documentation format that fits your workflow and style.

38 of 55

Advanced search techniques

38

Advanced techniques for core databases in the Search 101 - SR Database Cheat Sheet

39 of 55

Covidence Software

39

40 of 55

40

41 of 55

41

42 of 55

Covidence Settings

To use Covidence for solo projects, such as systematized reviews or narrative reviews, go to Review settings and change to 1 Reviewer.

42

43 of 55

AI in Literature Searching

GenAI for term generation
Rule-based tools for search translation
Searching with AI tools

43

44 of 55

GenAI for term generation

RAISE 3 Table 2

44

45 of 55

GenAI for term generation

45

46 of 55

GenAI for term generation

46

47 of 55

Rule-based tools for search translation

Polyglot Tool

Polyglot on the Evidence Review Accelerator (TERA) platform
Legacy access to Polyglot on Systematic Review Accelerator

Does a great job with Boolean operators and database syntax
Carefully review and update controlled vocabulary terms

47

48 of 55

Searching with AI tools

Tools:

ResearchRabbit: https://www.researchrabbit.ai/
CitationChaser: https://estech.shinyapps.io/citationchaser/
SciSpace: https://scispace.com/
Elicit: https://elicit.com/
And more!

Change is constant (especially in the AI landscape)!

Tools will come. Tools will go. Tools will change.

48

49 of 55

Searching with AI tools

Considerations:

The corpus - what sources are they searching?
Inputs - do you have the rights to upload or copy / paste the content?
Privacy

Are you required to register / sign up?
What information do they collect or reuse?

Costs - many AI tools have free and premium features, how does this impact your use?
Outputs - you are responsible for verifying the primary sources!!

Does the citation exist or is it a realistic looking fabrication?
Does the study data support the claim being made by the AI tool?

49

50 of 55

ResearchRabbit

50

51 of 55

Select "seed" articles

51

52 of 55

Deeper dive into an article

52

Paid features ↓

53 of 55

Article networks

53

54 of 55

Documentation & transparency

For each AI tool used, document the:

Model
Prompt(s)
Output(s)
Human oversight and decisions you implemented.

Disclose your use of AI tools as required by guidelines and principles of ethics and transparency in research.

54

55 of 55

Contact me with any questions or to schedule a consultation!

eginier@umich.edu

55