P-120 AI (Artificial Intelligence) based embryo assessment: What do embryologists think?

Abstract

Abstract Study question Do embryologists’ attitudes about AI assessment of embryos change over time and does perception and acceptance of AI differ among new and experienced/established embryologists. Summary answer Experienced clinical embryologists rely on AI technology less to assist with embryo selection than their more junior colleagues. What is known already While the objective efficacy of AI embryo assessment systems is being thoroughly scrutinized, and much work is underway assessing the theoretical attitudes of a variety of stakeholders toward such techniques, little information is available regarding the perception and acceptance of these systems by embryologists engaged in their routine clinical use. Previous research in the field indicates that junior embryologists generally have a higher level of trust in AI guided embryo assessment. To date, no studies have examined whether embryologists’ attitudes and perceptions change over time as the technology remains in routine clinical use. Study design, size, duration This is a qualitative study based on anonymous survey responses from clinical embryologists employed at a single large clinical laboratory. Embryologists who were present at the time of AI system introduction were invited to complete an initial survey in March 2023 (after 1 full year of use) and a follow-up survey in March 2024 (Group A). Embryologists who were not present at the time the system was introduced were surveyed in March 2024 (Group B). Participants/materials, setting, methods Anonymous surveys were conducted using the online survey platform SurveyPlanet (app.surveyplanet.com). Surveys consisted of eight questions relating to the embryo assessment AI platform Intelligent Data Analysis (iDA – Vitrolife) with answers recorded on a scale of 1-5, with 1 being the most negative response and 5 the most positive (n = 16). Statistical significance was tested using t-tests with the online software program Graphpad prism. Main results and the role of chance All embryologists (Groups A and B) reported ease of initial training in the use of the AI system (Mean score=1.5, SEM=0.27). Over time, embryologists perceived iDA to be increasingly time consuming (Increasing from 1.38 to 2.0), however this trend was not significant (p = 0.09). When posed the question of whether the embryologist would prefer to substitute the AI score for their own judgement the responses were overall not in favour of this move, with a mean of 2.67. This number did not differ significantly over time (p = 0.28) In the study laboratory a modified Gardner’s blastocyst grading system was used to rank embryos and select for utilization. In cycles where multiple good quality embryos of the same grade are available then AI could be utilised to differentiate among them and embryos with higher AI scores being selected preferentially. New scientists relied more on iDA to assist with embryo selection (mean = 52.14% of cases) compared to more experienced embryologists (27%) (p = 0.02). Overall embryologists’ showed trust in the implantation predictions of the AI (mean = 3.3). However, trust in AI was highest at implementation (Mean=3.63) and lowest following extended use (mean=3.0). This trend did not prove to be statistically significant (p = 0.11). Limitations, reasons for caution The trends observed in this study should be further verified by surveying a greater number of embryologists from various clinics as embryologists’ perceptions and opinions may differ at an individual or organization level. Wider implications of the findings This study is the first to examine how the perceptions and attitudes of embryologists towards AI change over time. This is important because embryologists ultimately decide the fate of embryos and their opinions influence how AI is utilized in a clinical setting. Trial registration number No

Ähnliche Arbeiten