Associations between the artificial intelligence scoring system and live birth outcomes in preimplantation genetic testing for aneuploidy cycles

Table 3 The embryonic and clinical outcomes of blastocysts with quartile AI scores

iDAScore	Group 1 (3.0–7.8, n = 107)	Group 2 (7.9–8.6, n = 136)	Group 3 (8.7– 8.9, n = 102)	Group 4 (9.0–9.5, n = 137)	Trend tests, p
KIDScore D5, mean ± SD	4.7 ± 1.6^abc	5.7 ± 1.5^ade	6.4 ± 0.9^bdf	7.4 ± 1.1^cef	< 0.001
Blastocyst morphological scores, mean ± SD	5.1 ± 1.2^abc	6.1 ± 1.4^ade	6.7 ± 1.1^bdf	7.4 ± 1.4^cef	< 0.001
D5 blastocyst, % (n)	23.4% (25)^abc	69.1% (94)^ade	98.0% (100)^bd	100% (137)^ce	< 0.001
Clinical pregnancy, % (n)	46.7% (50)^abc	74.3% (101)^a	62.7% (64)^bd	78.8% (108)^cd	< 0.001
Live birth, % (n)	34.6% (37)^abc	67.6% (92)^a	59.8% (61)^b	72.3% (99)^c	< 0.001
Pregnancy loss, % (n)	26.0% (13)^abc	8.9% (9)^a	4.7% (3)^b	8.3% (9)^c	0.007

The abbreviations “p”, “AI”, “D5”, SD, and “n” denoted p-value, artificial intelligence, day 5, standard deviation, and number, respectively. Trends tests were performed by analysis of variance or Cochran–Armitage test. ^a, ^b, ^c, ^d, ^e, ^f indicates the significant difference (p < 0.05) between groups by using the Mann-Whitney U test or the Fisher exact test

ISSN: 1477-7827