Merck backs startup bringing AI to fertility treatment


It's no secret that fertility treatment is expensive and has a low success rate, placing a big financial and emotional burden on would-be parents. Now, a Canadian startup aims to tackle that problem though the use of artificial intelligence.

Toronto-based Future Fertility has developed an AI called Violet that it says can predict the viability of cryopreserved eggs used for in vitro fertilisation (IVF), making sure that those most likely to result in a successful procedure.

The company has just come out of stealth mode with a first-round financing led M Ventures the venture capital arm of German drugmaker Merck KGaA – a major player in fertility treatment with drugs like Gonal-F (follitropin alfa) and Ovitrelle (choriogonadotropin alfa) – with Whitecap Venture Partners also taking part.

Violet has been trained to spot the features that predict IVF success based on thousands of egg images linked to anonymised health records, according to Future. It detects features of an egg that are invisible to the human eye but can predict the likelihood of fertilisation and embryo development.

As fertility starts to decline after age 30, increasing numbers of women are opting to freeze eggs as an insurance policy in case they choose to have children later in life.

A recent study found the number of cryopreservation cycles per year in the US went up almost 10-fold between 2010 and 2016, with a four-fold increase in Australia and New Zealand, and with the age at which women opt for the procedure trending down.

At an average cost of $20,000 per treatment cycle in the US, raising the chances of a successful pregnancy could make a huge difference to patients undergoing IVF.

At the moment it is estimated that only around a third of egg retrievals across all age groups result in live births in women aged under 35 when their eggs were frozen, falling to 16% in the over 35 age bracket.

Violet is already being used in a some fertility clinics around the world and according to the startup is consistently shown to be more accurate at predicting egg viability than current methods.

The startup calculates that the AI can predict successful fertilisation with an accuracy of 90% from a single egg image, building the odds in favour of a successful IVF procedure.

That's only the start of the process, however, and the AI less successful at predicting the success of the next stages in IVF, including survival of the embryo for the first few days after fertilisation, and its chances of implanting successfully in the womb.

That may improve with greater use of Violet as it rolls out in additional clinics, allowing further training of the AI, but Future says it is developing additional AI tools to address other aspects of fertility treatment.

Image by Marjon Besteman from Pixabay