How Can AI Help Me Understand if My Frozen Eggs Will Be Usable?
There is a lot of buzz around AI lately. From ChatGPT to Alexa to Netflix recommendations, it seems to be permeating many aspects of our personal and work lives. But did you know that AI is also gaining momentum in the fertility space?
How Can AI Help Me Understand if My Frozen Eggs Will Be Usable?
AI is able to process massive amounts of information, which means it’s perfect for solving complex challenges that humans haven’t been able to figure out.
Understanding a person’s egg quality is one of those challenges that AI is now able to give us insight into.
How can AI assess egg quality?
Before explaining what AI can do, let’s first start with what humans cannot do.
After your egg retrieval as part of your egg freezing process, an embryologist will examine your eggs under a microscope to strip them of their cumulus cells and identify the mature eggs. (Typically, only mature eggs are frozen.)
While examining the eggs, embryologists can see abnormalities in the way an egg looks compared to a “normal-looking” egg – these are called dysmorphisms. However, the occurrence of dysmorphisms is rare and scientific studies have not been able to consistently link these egg dysmorphisms to differences in how an embryo develops once those eggs have been fertilized.
As such, there is actually no objective or agreed-upon way today that embryologists can assess an egg’s quality by looking at it. This is where the AI comes in.
Embryologists can now take images of the freshly retrieved eggs while under the microscope and upload them to AI-powered software for a quality assessment. This software uses a specialized AI model that can perform image recognition to analyze every pixel within the image, identify patterns and generate a prediction about the egg’s reproductive potential.
That means that as an egg freezing patient, you would be able to understand the likelihood of each of your frozen eggs developing into a healthy embryo when it’s time to use them. One of the egg quality reports in the market today also provides a personalized prediction of your overall chances of having a baby from your frozen eggs.
Having personalized insights like these can help you – in consultation with your clinician – determine if you should do more freezing cycles to improve your overall chances of success, or if you should consider other paths and/or timelines to parenthood.
How does an AI model designed for egg quality assessment work?
Image classification models are used throughout the healthcare industry to analyze many different types of medical images (e.g., x-rays, MRIs) and generate predictions about health outcomes. In each case, the model needs to first be trained by analyzing lots of images where the health outcome is already known.
For example, an egg quality model will process each egg image in its training dataset and be provided with additional information about whether the egg formed a blastocyst or not. (A blastocyst is a day 5-6 embryo. Reaching this embryo stage is a good proxy for high egg quality because embryos at this stage have a higher chance of resulting in a baby.)
It will repeat this process tens of thousands of times with different egg images and their respective outcomes – enabling the model to learn patterns and detailed features within the image that indicate whether an egg is more likely to develop into a blastocyst.
Once the model has been trained, then it’s ready to start analyzing new egg images where the blastocyst outcome is not yet known (i.e., images of the eggs you have just retrieved). The model uses its previous learning to make a prediction about the new egg’s likelihood of developing into a blastocyst – based on all the other egg images and outcomes it analyzed during its training.
As you can imagine, it’s very important to ensure that the model’s training dataset includes massive amounts of high-quality images from a balanced distribution of different patient types to ensure that it provides accurate insights across a varied patient population. For example, the largest egg quality model available today was trained and tested using more than 100,000 egg images from eight different countries, across a wide range of patient ages.
In cases where a model has a small training dataset or low patient diversity, the model can develop a bias in its predictions towards certain patient types or outcomes. For example, if a model is only trained on eggs from older patients, then it might not be very good at predicting outcomes for younger patients who may have eggs and outcomes that look different. Or a model created with data from patients in only one country might not work well in other countries, as studies have shown differences in reproductive health data across different regions.
What benefits do I get from including personalized egg quality assessments in my treatment?
By understanding the quality of each of your own eggs through these AI-powered reports, you’re empowered with personalized information that can help you and your fertility team determine how to best approach your future treatment and parenthood options.
A common approach for estimating success rates from frozen eggs has historically been to use an online egg calculator. These calculators base their formulas on scientific studies that have analyzed egg freezing outcomes across different age groups. The calculators then enable you to enter your age and the number of eggs you froze to receive an estimate of your chances of having a baby.
But – as with any health situation – it's important to remember that everyone is different. While age is correlated with egg quality, considering age alone is not enough to predict the quality of an individual’s eggs.
For example, if you froze 15 of your eggs in one cycle in your late 20s, your clinician may assume that you have great egg quality – as general estimates of egg quality are based on age. With a high number of eggs retrieved, that would give you pretty strong odds of having a baby from just one freezing cycle.
However, a closer look at the AI’s personalized analysis of your egg images may show that you don’t fall in line with what statistics would predict and your egg quality may be lower than expected – in which case your chances of success could decrease significantly. This may lead you to consider undergoing another egg freezing cycle to feel more confident about your success rate when it comes time to use them.
Do all clinics do these AI-based egg quality assessments? How can I access them?
While this technology is currently available in more than 15 countries, it’s still relatively new and not every clinic offers it. In order for the AI to analyze your eggs, the image needs to be taken right after your retrieval in the fertility lab. If having these personalized insights is an important factor in your clinic selection process, be sure to ask the locations you’re considering if they employ AI-powered egg quality assessment tools in their lab.
The use of AI in fertility treatment is continuing to evolve. As with many aspects of our lives, the role AI can play in our day-to-day is just starting to emerge. So far, its ability to assess egg quality beyond traditional age-based estimates is proving to be a promising new development for egg freezers. While freezing your eggs isn’t guaranteed to result in a baby, leveraging AI-based reports can give you insight into your own personalized egg quality, helping you set clearer expectations for success and make more informed decisions with your clinic team along your fertility journey.
Medical Disclaimer:
The information provided in this blog is intended for general informational purposes only and should not be considered as a substitute for professional medical advice, diagnosis, or treatment. Always seek the advice of your healthcare provider or qualified medical professional with any questions you may have regarding a medical condition. Never disregard professional medical advice or delay in seeking it because of something you have read in this blog.
