“`html
Artificial Intelligence (AI) has the potential to assist doctors in more accurately identifying eggs during IVF procedures that are likely to result in pregnancy. This advancement exceeds the precision of current techniques.
During IVF, physicians utilize ultrasound imaging to track the growth of follicles—tiny cavities in the ovaries that house eggs—to determine the right moment for a hormone injection called the ‘trigger.’ This injection prepares the eggs for collection and fertilization with sperm, aiming to create viable embryos. Timing is critical because the injection is less effective if the follicles are either too small or overly large at that point. Once the eggs are harvested and fertilized, a selected embryo is then implanted into the uterus with hopes of achieving pregnancy.
In their study, researchers employed ‘Explainable AI’ methods—an AI approach that allows for transparency in its operations—to evaluate historical data from over 19,000 IVF patients. They examined which sizes of follicles correlated with higher success rates in retrieving mature eggs that could lead to births.
Their research indicated that administering the hormone injection when a significant portion of follicles was between 13-18mm in size was associated with a greater retrieval of mature eggs and subsequently a higher likelihood of successful pregnancies.
Currently, medical practitioners focus on measuring the largest follicles through ultrasound and typically administer the trigger injection when they observe two or three lead follicles that are larger than 17 or 18mm.
The researchers propose that focusing on increasing the number of moderately sized follicles could enhance the retrieval of mature eggs and boost birth rates.
The team believes their findings underline AI’s potential to tailor IVF treatments more effectively, ultimately improving patient outcomes and increasing the chances of bringing home a baby. They intend to develop an AI tool that integrates these insights to personalize IVF procedures and assist clinicians in making informed decisions throughout the process, with plans to seek funding for clinical trials.
The research, which has been published in Nature Communications, involves collaboration among scientists from Imperial College London, University of Glasgow, University of St Andrews, and medical professionals at Imperial College Healthcare NHS Trust. The work is financed by UK Research and Innovation and the National Institute for Health and Care Research (NIHR) Imperial Biomedical Research Centre (BRC).
Dr. Ali Abbara, an NIHR Clinician Scientist at Imperial College London and a Consultant in Reproductive Endocrinology at Imperial College Healthcare NHS Trust, who co-led the study, stated:
“IVF brings hope to many who struggle to conceive but it is an invasive, costly, and lengthy process. When it doesn’t succeed, it can be devastating, making it essential to maximize the treatment’s effectiveness.”
“AI offers a new way to approach IVF treatment, potentially leading to better patient results.”
“With IVF generating vast amounts of complex data, it can be challenging for doctors to leverage all this information when making treatment choices for their patients. Our study illustrates that AI techniques are well-suited to analyze this intricate IVF data. In the future, AI could provide precise recommendations to enhance treatment personalization, ensuring every couple has the highest chance of having a baby.”
Professor Waljit Dhillo, an NIHR Senior Investigator from the Department of Metabolism, Digestion and Reproduction at Imperial College London, Consultant Endocrinologist at Imperial College Healthcare NHS Trust, and co-lead author of the study, commented:
“Our discoveries could lead to new methodologies for optimizing IVF treatment success, resulting in more pregnancies and births.
“This is the first study to analyze a substantial dataset, proving that AI can more accurately identify follicle sizes most likely to yield mature eggs compared to existing methods.”
“This development is exciting; it suggests we can consider a broader range of follicle sizes to determine when to administer trigger injections rather than solely relying on the largest follicles, which is the current practice.”
Dr. Thomas Heinis, a co-lead author from the Department of Computing at Imperial College London, added:
“Explainable AI can be an invaluable asset in healthcare. Given the significant stakes involved in making optimal decisions, this method can aid doctors’ decision-making and enhance patient outcomes. Notably, we anticipate a significant increase in computing power, allowing us to base decisions on accurate data like never before.”
Infertility affects one in six couples, and IVF has become a critical option for patients seeking to conceive.
Trigger injection
In IVF treatment, a crucial decision involves timing the ‘trigger’ injection of hormones, such as human chorionic gonadotropin (hCG), to mature eggs for collection. How this injection is timed directly influences the number of mature eggs retrieved and the overall success of the treatment.
Medical professionals use ultrasound to evaluate the largest follicles. Generally, they administer the trigger injection when there are two or three lead follicles measuring over 17 or 18mm in diameter. However, this method lacks precision because it does not account for the size of each individual follicle or their probability of producing mature eggs.
Follicle sizes
In their retrospective analysis, the team applied AI techniques to data from 19,082 patients aged 18-49 who underwent treatment at one of 11 UK clinics, including IVF centers at Imperial College Healthcare NHS Trust, as well as at two clinics in Poland from 2005 to 2023. They investigated the sizes of individual follicles leading up to and including the day the trigger was administered.
The researchers discovered that follicles measuring 13-18mm were associated with an increased number of mature eggs being retrieved later. The data indicated that having a higher count of follicles within this size range on trigger day correlated with improved clinical results.
Additionally, they identified that overstimulating the ovaries—resulting in a larger number of follicles exceeding 18mm—on the day of the trigger could lead to an unwanted rise in progesterone levels. This spike can negatively affect IVF results by interfering with the development of the endometrium, the uterine lining critical for embryo implantation and subsequent pregnancy, thus reducing the likelihood of a successful live birth. The insights garnered from AI can guide the team in creating evidence-based IVF protocols informed by data, potentially enhancing treatment efficiency.
“`
