Clinical embryologist “A” is a specialist who selects the best sperm and eggs, and helps create and culture embryos suitable for pregnancy. Among multiple embryos, choosing the one most likely to lead to pregnancy is a crucial step—one that A traditionally performs by carefully examining embryos with the naked eye. While studying various embryo patterns and weighing options, A decided to get help from artificial intelligence (AI). After taking a photo of an embryo seen under a microscope and uploading it to a computer, the system returned a score in just one second—assigning an embryo grade and a pregnancy-likelihood score. (Because the product has not yet received sales approval from Korea’s Ministry of Food and Drug Safety, it is not currently being used in hospitals. This section describes a hypothetical scenario through the fictional figure “A” to illustrate how it could be used.)

In general, the IVF process works like this: in a single cycle, about 10 eggs are retrieved, fertilized with sperm, and then 5–10 embryos are cultured. From these, clinicians must select the “best” embryo to transfer into the uterus in order to improve the chance of pregnancy. Embryos are typically assessed visually by clinicians and clinical embryologists. However, because embryo patterns vary widely, judgments can differ from one professional to another, and variability between assessors can occur.

What if AI could learn what embryos with a higher likelihood of pregnancy look like—and then share that insight with clinicians? Kai Health’s embryo-analysis AI solution began with that question. The company’s AI solution, Vita Embryo, is designed to help clinicians choose embryos that are more likely to lead to pregnancy. In other words, once the program is installed, it functions like a “companion” that supports clinicians’ decision-making by helping them select embryos more effectively.

In an interview with ChosunBiz, Kai Health CEO Hyejun Lee said, “I decided to start the company while thinking about how infertility couples could have a child with less hardship,” adding, “Embryo selection can lack consistency depending on the clinician, so I felt we needed a more objective approach and higher accuracy.”

Dr. Lee is a physician-turned-entrepreneur who completed her residency and fellowship training at Seoul National University Hospital’s Department of Obstetrics and Gynecology, and later served as a director-level physician and head of an international clinic at Maria Fertility Hospital, before earning an MBA at Northwestern University.

Kai Health secures embryo-related data collected during infertility treatment from more than 10 medical institutions in Korea and abroad. Using those data, its AI learns which embryo appearances were associated with higher or lower pregnancy likelihood. As more data accumulate, selection accuracy is expected to improve. The core idea is to use AI algorithms—rather than visual assessment alone—to raise the hit rate of embryo selection.

Founded in October 2021, Kai Health has received recognition for its technology, attracting seed investments in 2022 from Smilegate Investment and Translink Investment, raising a cumulative total of KRW 1 billion.

Joo Eun-ji, a senior team lead at Smilegate Investment (which invested in Kai Health), said that because the company is still at an early stage, investors look closely at the founder’s background and the key team. She added that while there are startups with similar technologies in places like Australia and Israel, embryo-selection technology itself is still in an early phase—so there is no clear market leader yet—and that Kai Health is the only company developing an embryo-analysis AI solution for the Asian market.

She also noted that after the investment last year, Kai Health was selected for an AI training-data 구축 (data-building) initiative led by Korea’s Ministry of Science and ICT and the National Information Society Agency (NIA), which gave the company an opportunity to secure substantially more data than it could have collected on its own and to train its AI more effectively.

According to Dr. Lee, when clinical embryologists and clinicians culture and select embryos using conventional methods, selection accuracy typically remains around 40–50%. With Vita Embryo, she said, accuracy can increase to 60–70%. Here, “accuracy” refers to the proportion of embryos selected that are truly higher-quality embryos with a higher likelihood of pregnancy—it is not the same as an actual pregnancy success rate.

Dr. Lee added that one out of every ten babies born last year was conceived through infertility treatment, and that this proportion is expected to rise. In Korea, however, she noted that a single IVF cycle can cost more than KRW 3 million, while success rates remain around 30%. For couples who must repeat the process multiple times, this can be a significant emotional and financial burden.

Her view is that by improving embryo-selection accuracy and reducing the number of IVF attempts needed, Vita Embryo could help lower both time and cost burdens. Many couples undergo IVF three to six times; if selecting better embryos increases the chance of pregnancy, the number of attempts may decrease—and total costs could fall as well.

Kai Health has already obtained CE certification for its embryo-analysis AI technology. The company is currently pursuing clinical approval from Korea’s Ministry of Food and Drug Safety with the goal of entering the Asian market first. The company expects that it could obtain approval as early as the end of 2024 and begin sales in Korea.

Joo Eun-ji also said that because IVF can be expensive in the U.S. and Europe, there are cases where people travel to Asian countries such as Singapore and Thailand for fertility treatment—one reason Kai Health is pursuing clinical approval with Asia market entry in mind.

While the work began with the desire to help infertility couples, Dr. Lee acknowledged that building a startup is challenging day to day. She said that the business only becomes meaningful if many people actually use it, and it is difficult to know whether that will happen. She also expressed concern that, once the product is launched, the company will need to sell to hospitals via a B2B model—and hospitals can be conservative, potentially making adoption of new technologies difficult.

She expects that persuading clinical embryologists may also be challenging, since they currently select embryos by looking through microscopes themselves and may feel psychological resistance to relying on AI. Still, she emphasized that there is a global shortage of clinical embryologists, and an even smaller number of highly experienced senior embryologists—meaning the tool could be genuinely helpful.

Dr. Lee said she hopes the program will support better decisions when clinical embryologists and physicians select embryos together, and added that Kai Health ultimately aims to expand beyond infertility into contraception and pregnancy-related solutions—growing into a “family-tech” company that supports the creation of healthy families.

Source: ChosunBiz · By In-ah Lee · Published Jan 2, 2024 · Originally published in Korean · English translation

[VC Meets AI] Upload an Embryo Photo to AI and in One Second… “Pregnancy likelihood: 70%.”