P-254 Is Intelligent Data Analysis-score (IDA-score) a useful tool in daily routine IVF laboratory practice to select embryos? Preliminary results of an external independent series. (30th June 2022)

• Record Type: Journal Article
• Authors: Sarandi, S; Labrosse, J; Hammami, F; Arumugam, V; Bennani Smires, B; Puy, V; Vinolas, C; Cedrin-Durnerin, I; Peigne, M; Grynberg, M; Sifer, C
• Is Part Of: Human reproduction. Volume 37(2022)Supplement 1
• Publication Date: 2022-06-30
• Subjects: Human reproduction -- Periodicals; 618
• DOI: 10.1093/humrep/deac107.244
• Languages: English
• ISSNs: 0268-1161

Abstract

Abstract: Study question: Is IDA-score a useful tool to accurately classify embryos? Summary answer: IDA-score is a reliable tool to categorize embryos into top/good/poor morphological quality. However, hierarchical classification of embryos between IDA-score and embryologists was significantly different. What is known already: Since Preimplantation Genetic Test for aneuploidy is not authorized in France, morphological scoring by embryologists is the only way to select blastocysts for transfer at Day-5. However, morphological scoring cannot perfectly predict implantation likelihood. Promising recent artificial intelligence techniques might improve both hierarchical classification and clinical outcomes. For instance, IDA-score combined with Embryoscope (Vitrolife®) has been suggested as a way to objectively compare a given embryo to a large embryo dataset and generate a score that predicts the likelihood of implantation using a deep-learning approach. So far, no external independent analysis compares the accuracy of this score with human performance. Study design, size, duration: This retrospective-monocentric-study was performed in our reproductive medicine unit from September to December 2021. A total of 167 embryos from 58 couples undergoing ICSI were monitored on a daily-basis and scored at Day-5 by IDA-score and morphological scoring by a panel of senior-embryologists blinded from results of IDA-score. In all, 52 single embryos selected by embryologists were transferred during fresh or frozen cycles. Clinical pregnancy rates were defined by presence of fetal heartbeat on ultrasound. Participants/materials, setting, methods: Each embryo was monitored and scored at Day-5 by both morphological scoring and IDA-score. Only the best 3 embryos were ranked and compared. Morphological score was based on Gardner-and-Schoolcraft grading system and classified according to morphological score as top-quality (≥B4BB), good-quality (≥B3BB) and poor-quality (<B3BB;trophectoderm/inner-cell-mass:CC) embryos. IDA-scores were ranged from 1-10. We compared hierarchical classification by IDA-score and embryologists using appropriate statistical tests and IDA-scores of transferred blastocysts with successful implantation and those without. Main results and the role of chance: The mean age of women was 33.96±0.66 years. Morphological scoring led to 68.8% (115/167) top quality embryos, 80.2% (134/167) good quality embryos and 19.8% (33/167) poor quality embryos. Mean IDA-scores of embryos were respectively 9.06 ± 0.48 for top-quality, 8.72 ± 0.62 for good-quality and 7.12 ± 1.34 for poor-quality blastocysts (global p <10 -4 ). However, concerning hierarchical classification, there was a significant lack of global accuracy between IDA-scores and embryologists regarding the 3 best embryos of the cohort (68.3%; p <10 -4 ). Similarly, there was a significant lack of accuracy to classify embryos belonging to the same quality group (67.8% (78/115) for top-quality; 64.9% (87/134) for good-quality; 81.8% (27/33) for poor-quality). We obtained a clinical pregnancy rate of 38.4% (20/52) per transfer. We failed to observe any statistically significant difference between embryos transferred that implanted (8.92 ± 0.79) and embryos that did not implant (8.69 ± 0.93; p = 0.37) according to IDA-scores. However, IDA-scores tended to be higher in case of clinical pregnancy. In order for the observed difference of 0.2 in IDA-scores to reach the level of significance, at least 674 embryos should be transferred. Limitations, reasons for caution: These results should be interpreted with caution due to the retrospective design of the study and small effectives. Wider implications of the findings: We present preliminary results of an external independent series to confirm the efficiency of IDA-score in daily routine practice. Since IDA-score is not consistent with human scoring concerning hierarchical classification, its performance has to be further assessed to clearly establish the input of artificial intelligence in IVF laboratory practice. Trial registration number: not applicable