Artificial Intelligence in IVF

There has been mounting interest in the future of fertility treatment. Consultant Clinical Embryologist Andrew Thomson details how artificial intelligence fits the mould when it comes to improving the efficiency of fertility treatment.

Quick facts:

• Artificial intelligence (AI) is helping revolutionise fertility treatment by improving the efficiency of embryo grading.
• AI will rank the order of embryos to be replaced to the patient based on data from hundreds of thousands of previously-used embryos.
• This innovative process uses the same technology as face recognition though research is still growing in this area.
• In the future, (AI) may also be able to detect chromosome abnormalities in embryos

When discussing AI, traditional robots may come to mind. But when scientists like myself are talking about artificial intelligence in fertility treatment, particularly IVF, we’re referring to embryo analysis by advanced computers.

As an experienced embryologist and laboratory manager, I specialise in this ever-growing field.

AI can be quite difficult to wrap your head around. So, I’ve answered some typical questions about exactly how it works as an add-on to IVF.

What is AI?

Artificial intelligence in IVF is used as an extension to the time lapse monitoring incubators (TLM) most IVF labs now use.

TLM incubators take images of embryos every 10 minutes. This allows embryologists to observe every stage of development without disturbing their environment. Tens of thousands of anonymised videos of developing embryos are made each year.

AI in fertility treatment is still a growing subject area. Several research groups from around the world have been developing mathematical models and algorithms to improve the selection of embryos by analysing the data of those which have gone on to achieve successful pregnancies.

This data is then fed into a neural network for computers to assess. During this process, these clever computers use the same technology that identifies faces, animals, and objects.

AI has been utilised in other scientific and medical settings such as making quicker and more accurate diagnoses of breast cancer. The capability of AI in IVF is just the next step.

So, how exactly does it work?

• There’s a lot going on in an embryo that’s invisible to the human eye, but might not be to a computer.
• AI assesses each of the embryos for a patient and, as discussed above, compares these against ones which it knows have gone on to achieve pregnancies.
• It will also select against embryos which have developed in an abnormal way, for example by dividing directly from one cell into three cells, rather than into two.
• At the point of embryo transfer, AI gives each embryo a score (normally out of 10) and ranks them in the order each embryo should be replaced to the patient. This process is called embryo grading.
• Without AI, the grade of embryo may actually change each time the same embryologist looks at it. AI removes this subjectivity and makes more consistent decisions.
• In the future, artificial intelligence may also be able to detect chromosome abnormalities in embryos by simply watching how they develop. This is opposed to it going under more invasive genetic testing.

Can AI select the wrong embryo?

The order of the embryos ranked by AI may sometimes be inaccurate, but this is rare.

In all cases, embryos undergo a secondary review and selection process by a highly-trained embryologist, who can override any decision AI makes by using their own independent decision.

An embryologist will always be responsible for replacing a patient’s embryos, not a robot.

Who might be recommended for this type of screening?

AI is available to most IVF patients who have decided to use TLM incubators as part of their treatment.

It’s important to note that not all clinics will use AI as it is a fairly new feature, so you’ll have to ask your fertility centre directly if they offer the add-on.

What are the benefits of using AI for the patient?

By ranking the embryos based on AI algorithms, scientists hope it will speed up time to conception by transferring the best embryos first.

TLM incubation is still amber-lighted by the UK’s IVF regulator Human Fertilisation and Embryology Authority (HFEA).

The traffic light system is used to rate how effective an add-on is in improving the patient’s chances of having a baby.

This means that while data is promising, more research is needed for AI’s use in fertility to achieve the green light rating.

AI is so new that it doesn’t even have a score yet. But in my opinion, TLM incubation and AI should really be considered as one and the same. AI may end up simply changing the order in which we select embryos for transfer and freezing. 

For more information about the traffic light system, please visit HFEA’s website.

Is there a risk of overselling specific treatments?

AI is mainly just a ranking system so it won’t improve your egg or embryo quality. In order words, it’s just one piece of the jigsaw in helping a patient get pregnant.

There are also many other factors to consider in this process, such as:

• age
• lifestyle
• medical history.

Is it NHS-funded, private or both?

Most clinics will charge for AI as part of their TLM add-on – which will vary clinic to clinic.

The TLM incubators are the most expensive piece of equipment we have in the lab. It comes in at over £100,000 per incubator.

As it has not been proven beyond doubt that TLM improves outcomes, most clinics pass on this cost to patients with some charging up to £1000 per cycle. Others may offer it for free.

However, at my clinic, the Centre for Reproduction and Gynaecology Wales (CRGW), analysis of our lab’s key performance indicators demonstrate significantly improved outcomes when using time lapse compared to conventional incubators.

I’ve also presented TLM data for another IVF centre showing the same conclusions at the Fertility 2020 conference.

One of these indicators include significantly more blastocysts frozen for later use. A blastocyst refers to when an egg becomes a hollow ball made up of hundreds of cells. This is five or six days after fertilisation.

This means that patients will be able to undergo more embryo transfers from one egg collection and have a potentially higher cumulative pregnancy and live birth rate. Based on this, CRGW offer it for free.

 
What do you think about AI and its use in fertility treatment?

I’m a big believer in using TLM and have been involved in many AI studies.

I can see the benefit of AI when you have multiple good quality blastocysts to choose between. But if you have low egg and embryo numbers, then it probably won't be as useful to you.

Either way, as a patient, you should always be able to decide what is best for you. You should have the time to do your own independent research before opting for any particular treatment option.

If you’d like to know more about fertility treatment and your options, please contact your Peppy practitioner.

About the author

• Consultant Clinical Embryologist Andrew Thomson joined the Centre for Reproduction and Gynaecology Wales (CRGW) in December 2021 as a Consultant Clinical Embryologist and Laboratory Manager. He previously worked in some of the top performing laboratories in the UK.
• He is an internationally recognised scientist who has presented his research both nationally and internationally on multiple occasions.
• Andrew lectures at the University of Bristol each year. He's also lectured at multiple prestigious universities including the University of Oxford.
• Andrew is one of just 14 Reproductive Scientists to have attained Fellowship of the Royal College of Pathologists. This is the highest accolade attainable in the UK.
• He is actively involved in training reproductive scientists to the highest standards as an assessor for the Royal College and the National School of Healthcare Science.