SFEBES2022 Poster Presentations Reproductive Endocrinology (36 abstracts)
Follicle Sizes That are Most Likely to Yield Oocytes During In Vitro Fertilisation (IVF) Treatment
1Imperial College London, London, United Kingdom; 2Imperial College London-Section of Endocrinology and Investigative Medicine, London, United Kingdom; 3Imperial College Healthcare NHS Trust-Department of Endocrinology, London, United Kingdom; 4UKRI Centre for Doctoral Training in AI for Healthcare, Imperial College London, London, United Kingdom; 5University of St Andrews-Department of Data Science, Scotland, United Kingdom; 6Imperial College Healthcare NHS Trust-Department of Reproductive Medicine, London, United Kingdom
Background: Infertility affects 1 in 6 couples causing devastating psychological impact. In vitro fertilisation (IVF) treatment can aid couples to conceive, but personalisation of treatment is needed to optimise patient outcomes. Machine learning can aid in the analysis of large complex datasets such as those encountered during IVF treatment. One example is in determining the optimal follicle size on the day of trigger to maximise the number of oocytes collected. Both follicles that are too small and those that are too large are less likely to yield oocytes. Therefore, determining the optimal follicle size range for oocyte yield provides valuable information for the clinician during cycle management.
Method: Retrospective cohort study analysing follicle sizes on the day of trigger compared to the number of oocytes retrieved from 8,030 IVF cycles conducted at the Hammersmith IVF unit in women aged under 35 years. Random forest ensemble machine learning technique was used to identify the follicle sizes that most contributed to the number of oocytes retrieved.
Results: Follicles of 11-19 mm on the day of trigger were most likely to yield oocytes, and mature oocytes, following both human chorionic gonadotrophin or gonadotrophin releasing hormone agonist. The number of follicles in the 11-19 mm range was more closely associated with the number of oocytes retrieved (r2=0.58) than follicles smaller or larger than this range (r2=0.051). The most predictive follicles sizes on the day prior to trigger (Day -1) were 10-18 mm, and the day prior to that (Day -2) were 6-17 mm, consistent with the median follicle growth rate of 1-2 mm per day.
Conclusion: Follicles of size 11-19 mm on the day of trigger are most likely to yield oocytes, providing vital information for clinicians managing ovarian stimulation to optimise patient outcomes.