Revolutionizing IVF: How AI is Transforming Success Rates

Research funded by NIHR has unveiled that artificial intelligence (AI) could assist physicians in enhancing the success rates of in-vitro fertilisation (IVF) procedures.

The study revealed that AI might aid in identifying the optimal time to target follicles – tiny sacs in the ovaries that contain eggs – thus boosting the chances of successful IVF outcomes.

The investigation was financed by the NIHR Imperial Biomedical Research Centre and UK Research and Innovation, led by scholars from Imperial College London, University of Glasgow, and University of St Andrews, along with clinicians from Imperial College Healthcare NHS Trust. It has been published in Nature Communications.

Understanding IVF

Infertility impacts one in six individuals and couples, often motivating them to pursue IVF treatment for conception. Current success rates for IVF are heavily influenced by the age of the women receiving treatment, with the highest pregnancy rates observed among patients aged 18-34 at 42%.

In IVF procedures, physicians utilize ultrasound imaging to track the size of follicles to determine the appropriate moment for administering a hormone injection known as the ‘trigger’. This process readies the eggs for harvesting and ensures they are prepared for fertilisation with sperm to form embryos.

The timing of the trigger administration is vital, as it proves less effective if the follicles are either too small or too large. After the eggs are harvested and fertilised, an embryo is chosen and transferred into the uterus, which hopefully results in pregnancy.

Employing AI for customising IVF treatments

This research employed ‘Explainable AI’ methods – a variety of AI that enables human understanding of its processes – to evaluate historical data from over 19,000 patients who had undergone IVF treatment. The researchers investigated which follicle sizes were linked to higher rates of retrieving mature eggs leading to live births.

Currently, medical practitioners determine the largest follicles using ultrasound scans and administer the trigger injection when two to three follicles are over 17-18mm. However, their findings imply that optimising the ratio of medium-sized follicles targeted with the hormone injection (ranging from 13mm to 18mm) could enhance egg retrieval and improve birth rates.

Professor Waljit Dhillo, co-senior author of the research, Dean of the NIHR Academy, and NIHR Scientific Director for Research Capacity and Capabilities, stated: “Our study is the first to evaluate a large dataset demonstrating that AI can pinpoint the specific follicle sizes most likely to yield mature eggs more accurately than existing methods.”

“This is an encouraging advancement as the results suggest we can leverage data from a broader array of follicle sizes to determine when to administer trigger shots to patients, as opposed to solely relying on the size of the largest follicles – which is the current clinical approach.”

Dr Ali Abbara, NIHR Clinician Scientist at Imperial College London and Consultant in Reproductive Endocrinology at Imperial College Healthcare NHS Trust, as well as co-senior author of the study, remarked: “In the future, AI could be utilised to offer accurate guidelines for enhancing decision-making and assist in tailoring treatments, ensuring each couple has the best possible opportunity to conceive.”

The research team intends to develop an AI tool that will apply insights from their study to personalise IVF therapies. This tool may support clinicians’ decision-making at various stages of the IVF process. They plan to seek funding to advance this tool into clinical trials.