Vetnews | Julie 2025 10 « BACK TO CONTENTS Leading Article The themes and noun phrases identified using auto-coding were reviewed manually (SSG) and refined by merging, moving, or deleting themes and noun phrases that were thought to be incorrect when interpreting them in context. Only a small proportion of the auto-coded themes and noun phrases were deleted, many of which were related to a study’s methodology for example, themes such as “Linear regression” or “Bayesian analysis” were removed. Other auto-coded themes were deleted if the theme was identified in less than 5% of the papers and could not be merged with another theme. Merging was more commonly performed on noun phrases and only occurred if two themes or noun phrases could be linked by a common idea or could be considered one and the same. For example, the noun phrases “Public health” and “Human health” were merged to form a single noun phrase. Rarely was a new parent theme created manually with the exception of the “Species” theme which has “Moose (Alces alces)” and “White-tailed deer (Odocoileus virginianus)” nested as noun phrases. It was thought important to keep these noun phrases as they were found in high frequency throughout the papers related to climate change impacts on vector distribution however, they did not fit into any existing parent theme. Noun phrases were also moved from one parent theme to another. For example, the AI auto-coding placed the noun phrase “Habitat management and protection” under the “Habitat” theme before it was manually moved under the “Management” theme. It could be argued that this noun phrase could belong to both parent themes, however, “Management” was selected based on the interpretation of the text segments that had been assigned during the autocoding process. In the end, very few edits were made to the autocoded parent themes with most of the manual corrections centred around the movement of noun phrases. Results Search Results The literature search yielded 2,249 citations (after removing duplicates) of which only 372 (16.5%; 372/2,249) met the inclusion criteria after being screened by all three reviewers. Of these 259 (69.6%; 259/372) contributed to the analysis focused on climate change and wildlife health while 113 citations (30.4%; 113/372) made up the vector distribution subset (Fig. 1). The datasets analysed during the current study are available in the Scholarly Commons: The University of Pennsylvania’s open-access institutional repository; https:// repos itory. upenn. edu/ handle/ 20. 500. 14332/ 60593. The number of citations retrieved varied across each year (Fig. 2) with results from the linear regression model suggesting that approximately 60% of the variance in the number of citations related to climate change and wildlife health could be accounted for by yearly changes (R2 = 0.637; p-value = 0.0001). In comparison, yearly changes accounted for approximately 50% of the variance in the number of citations related to climate change and vector distribution (R2 = 0.498; p-value = 0.0016). AI‑based thematic analysis—climate change and wildlife health A total of 17 parent themes and 103 noun phrases were identified across the 259 citations focused on climate change and wildlife health (Table 2 and Supplementary Table 1). The top noun phrases identified across all the papers focused on concepts related to Figure 1: Number of records yielded from the scoping review across five databases: Scopus, Web of Science, Zoological Records, ProQuest, and CAB Abstracts, and included in the thematic analyses
RkJQdWJsaXNoZXIy OTc5MDU=