Page 11 - AIH-2-3

Artificial Intelligence in Health AI in embryo selection for ART

Figure 2. Flow diagram illustrating the study selection process in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews
and Meta-Analyses
Abbreviation: AI: Artificial intelligence.

Table 2. Bibliometric characteristics of the included studies well-designed, and externally validated studies to confirm
the reliability and generalizability of AI models in the
Characteristics Corresponding entities Numbers Percentages clinical environment.
Publication year 2015 – 2018 1 3.33
2019 – 2021 13 43.33 3.3. Bibliometric characteristics of the included
2022 – 2024 16 53.33 studies
Publication type Article 23 76.67 This systematic study suggests that incorporating
Review 5 16.67 engineering principles into the evaluation of online
Conference proceeding 2 6.67 databases can enhance reliability and reduce publication
Quality of paper Q1 16 53.33 bias. Three researchers carefully reviewed the titles
and abstracts of the identified studies to reduce bias.
Q2 10 33.33 In addition, an experienced academic expert reviewed
Q3 1 3.33 the work to identify and address any potential
inconsistencies or biases. As shown in Table 2, the
selected publication types include 23 articles (77%), five
studies by Glatstein et al. and Salih et al., outcome review papers (17%), and two conference proceedings
21
20
assessors were frequently not blinded, increasing the risk (6%). In terms of geographical distribution, the
of subjective determination regarding embryo viability United Kingdom leads with 14 publications, followed by
and implantation success. Moreover, selective reporting the USA (12), Switzerland (2), Japan (1), and Bosnia and
was also an issue, with some studies failing to disclose Herzegovina (1).
limitations and adverse effects. For example, while Tian
et al. acknowledged the necessity for external validation, 3.4. Appraisal of the study quality
22
this aspect was not consistently addressed across other This systematic study suggests that incorporating
studies. Although the current review underscores engineering principles into online database searches can
promising progress in AI-driven embryo selection, the enhance reliability and reduce publication bias. Three
identified RoB highlights the need for future prospective, researchers carefully reviewed the titles and abstracts of the

Volume 2 Issue 3 (2025) 5 https://doi.org/10.36922/aih.7170