The research-to-practice gap.
Today’s rapidly changing health care settings require medical and nursing professionals and students to remain up to date on trending research, topics, and evidence for guiding practice. While this may sound fundamental for nurses, multiple barriers make this incredibly challenging. Factors such as limited time, large volumes of new research to sift through, and experience with reading and analyzing research contribute to what is known as the research-to-practice gap. This blog post will explore how to harness AI and technology to gain a high-level and comprehensive overview of a research topic of interest.
Define the topic.
Before leveraging AI tools, it’s critical to develop the focus of the topic of interest. It is helpful to frame or organize your topic or area of interest to ensure the search is thorough. For example, you could use the PICO format (patient/population, intervention, comparison, and outcomes) to phrase your question or area of interest.
Let’s say you want to learn more about skin damage related to external urinary devices for adult females. A good PICO question might be: Among adult females in acute care settings, what type of skin damage occurs when using external urinary devices compared to those who do not use these devices?
Search the literature.
Searching academic databases can be daunting, and a rigorous search often requires the help of a medical librarian. A large language model (LLM) such as ChatGPT can help you find a comprehensive search strategy. Try prompting that LLM by asking the following, replacing the particular topic example with your own well-developed PICO question:
“Please devise a search strategy (using MeSH and title/abstract keywords) for the following PICO question: what type of skin damage occurs when using external urinary devices compared to those who do not use these devices? Provide Boolean phrases and the PubMed syntax.”
Below is a screenshot of a portion of the ChatGPT search strategy output in response to a different PICO question having to do with bullying in nursing:
This response can in turn be copied and pasted into the PubMed search builder, as seen below:
And below is the syntax copied into the PubMed database to search for relevant articles. Only the first article in the PubMed results is shown in the screenshot.
Is this method reliable enough?
While ChatGPT and other chatbots are known to be unreliable for finding the best peer-reviewed academic literature because of a tendency to “hallucinate” articles that don’t exist, in my experience, ChatGPT is quite precise in the focused task of helping design search strategies and outputting database syntax that can provide tightly replicable literature searches. However, it should be understood as a tool; critical thinking remains essential.
Leverage AI tools to understand and develop the topic.
Now that you have a well-developed topic question and search procedure, your next step is understanding the literature to inform your practice. There are various tools that nurses can use to make this process less daunting.
Research Rabbit is a free website that allows users to upload one (or more) papers of interest. The system will automatically provide the user with a map of similar authors on the same topic and with connected papers (authors who cite each other). This allows users to gain a complete overview of the history and ongoing research development on a specific topic.
Voyant Tools, Elicit, and SciSpace are free, AI-powered tools that allow users to upload articles (located during the search process), providing users with an AI-generated paper analysis. When trying to get a broad overview of a topic or area of interest, summarizing the article’s main points can be helpful and a timesaver. Most of these open-access tools allow users to download the data extraction into Excel tables. Nurses can then take these tables to research councils or unit meetings to help guide practice using current research.
Any time nurses are looking at research before deciding to change practice in ways that can affect the end users, in this case patients, great and detailed care must be undertaken. Given the unreliability of AI-powered tools, nurses should work together to confirm all data extracted from articles, as chatbots like ChatGPT still tend to ‘hallucinate’ (or make up) information. Careful examination is required before making patient-level practice changes.
Incorporating AI and technology into your research process can give you a high-level and comprehensive overview of your chosen topic. As a nurse, you can enhance your research skills and help guide the application of unit-level and practice-specific evidence in clinical settings. By embracing the power of these technological advancements while maintaining a healthy caution, you can confidently and efficiently navigate the complex world of nursing research and evidence-based practice.
By Justin Fontenot DNP, RN, NEA-BC, FAADN, associate professor, Tulane University School of Medicine, Program of Nursing. His previous post on AI and writing and research strategies was “Can Nursing Students Benefit From Using Artificial Intelligence?”