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.

– Victoria Perweiler, Jonathan Friedlander, PhD & Geoffrey W. Smith



First Five
First Five is our list of essential media for the month. For our full list of interesting media in health and science, updated regularly, follow us on Twitter or Instagram.

1/ Trust
Pandemics and political turmoil are very bad for trust as evidenced in the 2021 Edelman Trust Barometer.

2/ Teeth
It has been tough to get to the dentist during COVID, this piece in Scientific American explains why for modern humans good oral care is a must.

3/ Tools
Thomas Kuhn posited that scientific progress was made in fits and starts marked by periodic “paradigm shifts” that upset the status quo in a non-linear fashion. Peter Galison has emphasized the impact of tools and methods as being of paramount importance in creating the needed conditions for these disruptions to occur and encouraged the notion that technology creates the tangible breach or disruption of a field. This Science paper provides a good example of how new tools and methods come together to create new insight, in this case into understanding and predicting the path of metastatic cancer cells. This interactive piece in The New York Times Magazine explores the intersection of COVID and ultrafast / ultracheap genome sequencing and how it might offer up the next paradigm shift in healthcare.

4/ Transmission
Work out of McGill University published in the journal Developmental Cell has meaningfully advanced our understanding of epigenetics by explaining how environmental information is transmitted by non-DNA molecules in sperm. In particular,  the study identifies a non-DNA based means by which sperm remember a father's environment (diet) and transmit that information to the embryo. Nature and nurture (of the parent) seems to be the answer…

5/ And Digitalis
Finally, we can’t resist the opportunity to point out our namesake in the science news — as reported in the Journal of Ecology, it turns out that when the common foxglove (Digitalis purpurea) came to the Americas, it evolved the shape of its flowers to support pollination by hummingbirds preferentially over bees. Proving sometimes it's the birds or the bees.



Digitalis Commons
Public-Interest Technologies for Better Health

Digitalis Commons is a non-profit that partners with groups and individuals striving to address complex health problems by building public-interest technology solutions that are frontier-advancing, open-access, and scalable.

The Digital Public Good Alliance put out an interesting piece on Financing the Digital Public Goods Ecosystem that is worth a read this month.

‍