A Fertile Ground for AI Companies
Last month, we wrote about the digitization of healthcare and how it paved the way for new types of companies. One such type is companies tackling rising infertility rates with a mix of machine learning and direct-to-consumer products. What is striking is the sheer number of companies doing almost exactly the same thing in this area. Why do so many companies come out with such undifferentiated propositions? Can they all succeed?
First, a few words about the state of fertility. Per a 2020 report from the US Center for Disease Control and Prevention, birth rates in the US are at a historic low. The general fertility rate is currently less than half of what it was at the peak of the baby boom, and at 1.71 live births per woman is well below the replacement rate of 2.1. However, these aggregate trends in fertility mask large heterogeneity across demographic groups. The trends have been driven mostly by cultural and societal changes pushing couples to wait longer and have fewer children: birth rates have fallen for women in their 20s and early 30s but have risen by an average of 3% annually since 1985 for women in their early 40s. These trends come at a cost including increased preterm birth rates and a painful reality for many — it takes longer to get pregnant than ever before. While it is clear that the risk of miscarriage and chromosomal abnormalities begins to rise at 35 and goes up sharply for women 40 and older, the causes of male infertility should not be overlooked. Between lower sperm counts and a more advanced age at conception, a third to half of fertility struggles are caused by male infertility. It is in this evolving context that the fertility industry and its main tool in vitro fertilization (IVF) have seen steady growth.
Among the most promising clinical applications of artificial intelligence is its use in IVF and other fertility treatments. The pace of progress in the field has accelerated in the last few years and access to cheap computing power and commoditized algorithms has democratized research and entrepreneurship. There is a myriad of new AI-driven companies recently created and financed which cover the entire fertility journey: from at-home sperm testing kits analyzing the common measures of count, motility and morphology (Dadi, Legacy, Exseed), to finding predictive patterns in patients’ lifestyles (Inanna, Univfy), to selecting embryos with the best chances of live birth (Overture Life, Ovation, Presagen, Embryonics, Future Fertility, Stork ai). This list does not even include all the apps supposedly helping couples get pregnant via menstrual cycle tracking. When so many players enter a market still restricted in size (about 3,000 clinics in directly accessible markets, serving roughly 1.3 million cycles annually), success often relies not only on outcomes but on the ability to strike partnerships for faster scaling.
The high predictive value that these statistical machines are able to produce often lack true biological understanding of fertility overall. While machine learning may deliver clinical benefits, the work to understand mechanistic insights is key to not only predict but prevent and treat underlying conditions. We hope that all these innovations will serve couples trying to conceive and lower the cost barriers to reproductive technologies, but recognize that the real impact of the financing gold rush in this area is yet to be seen.