A Japanese research team has developed artificial intelligence models that can help deliver personalized infertility treatments by predicting the number and quality of eggs that a woman has.

The AI models can accurately predict a patient’s ovarian function by using answers to simple questions — such as age and menstrual cycle — and a small sampling of blood, according to the group study led by University of Tokyo researcher Miyuki Harada.

The group collected data from 442 patients who were undergoing infertility treatment between June 2021 and January 2023 in Japan. Using their medical records and residual serum, the group created multiple AI prediction models.

In predicting the number of eggs, the model that evaluated five factors, including the value of anti-Mullerian hormone, was most accurate. On egg quality, the model that analyzed 14 factors from the patient's medical records and residual serum analysis had the highest accuracy.

Both models were more accurate than measurements given by the standard hormonal assessment method.

Until now, accurately assessing the health of an ovary was difficult due to individual differences. But with the AI models, ovarian function can be measured more easily and appropriate infertility treatments can be recommended to suit each person, the study said.

The models can also help with preconception care, such as by giving advice on lifestyle adjustments for future pregnancies in mind, it said.

The study comes at a time when patients experiencing infertility are on the rise, as a growing number of women choose to marry and have children later in life.

According to the health ministry’s most recent numbers, in 2023, the average age of a Japanese woman giving birth to her first child was 31.0, a significant increase from 1970 when the average age was 25.6.

In 2022, around 540,000 patients used some form of assisted reproductive technology, which led to 77,276 births, or around 1 in 10 children born in Japan.