To what extent should technology be involved in reproduction? Although AI can offer some benefits in

fertility, according to Julian Koplin, Ph.D., a member of the Monash Bioethics Society, “it does involve

AI interfering with a particularly sensitive area of human life, which needs to be handled carefully and

respectfully.” This illuminates the important bioethical dilemma that can often face the incorporation of

AI into IVF and fields of reproduction [1]. While AI is makingstrides in improving optimization of IVF

procedures, there continues to be hesitations in its realistic applicability and ethical principles.

AI is a broader term referring to the use of machines with intelligence [2]. Machine learning falls within

this umbrella and allows machines to adapt to experience without explicit programming [2]. Deep

learning is a subset of machine learning that uses artificial neural networks to predict or learn from

information and examples [2]. With AI becoming more mainstream, its incorporation into medicine is

being increasingly explored. One such example is its use in in vitro fertilization, or the process of IVF.

IVF is an expensive process that allows for the fertilization of an egg with a sperm outside of the uterus,

and subsequent implantation of the fertilized embryo into the body for pregnancy. The current success

rate of IVF is 30%, but AI can possibly improve its efficacy [2].

AI use in IVF allows for personalized predictive models to help tailor the dose of gonadotropins

(hormones to stimulate follicle production) given to women using markers like age, BMI, follicle count

(which increases the number of available eggs for release), and hormone levels including FSH, estrogen,

hMG, progesterone, and LH which play a role in regulating the menstrual cycle [3, 4]. Gonadotropins are

administered to stimulate the ovaries to produce multiple follicles to produce more eggs which can be

extracted for fertilization [4].

Predictive models can be used to optimize the number of oocytes, or eggs, as well as choose the best day

in a woman’s cycle to administer hormones, as well as integration with previous success data from other

IVF attempts or cases [3]. Ovarian stimulation protocols allow for more eggs to be released than just one

during ovulation or the release of an egg [4]. AI can optimize and personalize the ovarian stimulation

protocols based on individual patient characteristics and histories [3].

AI also offers measures that eliminate human error for gamete selection in both sperm and eggs to

increase the success of IVF. For sperm evaluation, AI can assess the gamete motility, morphology (shape),

and various movements like velocity, linearity, and amplitude of lateral head displacement (a measure of

movement and motility using the sperm head) using high resolution video [3]. Oocyte quality is assessed

under a microscope and can be subject to human error [3]. With AI, oocyte quality can be measured using

image analysis of morphological criteria of different parts of the oocyte [3]. In one study published in

2023, AI predicted oocyte fertilization with 82.4% accuracy, and was 75% accurate in predicting live

newborn birth [5]. With this increased gamete selection quality, there will likely be an increase in

implantation success and rates of pregnancy from IVF.

The existing companies using IVF and AI integration include BELA which assesses for chromosomal

abnormalities [6], and AIVF’s EMA. EMA is used for embryo evaluation using metrics like “visual

quality, morphokinetics and ploidy probability for every embryo.” [7] They evaluate early stage embryos

and blastocysts using predictive models and morphology assessments to establish which embryo will have

the best chance of success [7].

Despite these benefits, there are challenges with this technology’s implementation. Its current use has

great potential but currently, very small data sets from which the AI is developed and learned from [3].

There is a lack of standard procedure from how data is collected and reported which can impact the

generalizability of its performance [3]. If the data from which the AI is developed is not a large enough

sample, it can introduce bias into how predictions are made. Beyond this, there are increasing hesitations

for the input of sensitive data into these internet-based models, which can leave patients vulnerable to data

leaks [3].

One manifestation of ethical concerns related to AI use in IVF is the exacerbation of medical racism.

These limited sample sizes that are collected from single-centers could further the medical racism that

already fails to put people of color and minorities into existing studies. Medical racism manifests when

women of color are dismissed in their concerns–which also includes reporting and seeking help for

infertility, making them less likely to receive quality support or inclusion in studies given social stigmas

of “hyperfertility” [8]. Without the emphasis or desire to ensure that these predictive models are inclusive,

they could continue the trend that fails to prioritize the varying health needs of women of color.

For example, in studies as recent as 2022, some cardiovascular trials consisted of only 3.2% Black women

participants and oncology studies had 2-5% Black women represented. [9] Therefore, when making

predictive models for IVF, it is imperative that there be larger movements towards including more diverse

groups of women to move towards widespread implantation and pregnancy success rates.

Additionally, an argument could be made for the lack of access to such AI-based technologies. IVF is

already an expensive procedure catering to the middle and upper white class. On average, IVF costs

$15,000 per round [4], and a 2021 study conducted by Emory University found that non-Hispanic Black

and Hispanic women were 70% less likely to have infertility treatment versus White women [8]. These

technologies are already largely not accessible for low-income minorities which hinders their ability to

exercise their reproductive freedom based on financial constraints. Even with IVF, these populations may

still be excluded from access, and may miss out on the ability for early genetic screens and maximizing

the selections of their embryo which could hinder the future health of their children from the start. This

represents capitalist exclusion of groups of peoples whose wellbeing has not been put at the forefront of

technological innovation. Women of color continue to be excluded from spaces of quality healthcare and

inclusive representation within clinical trials and testing procedures.

In this case, the use of AI represents a massive stride in eliminating human error and maximizing the

effectiveness of IVF procedures. This would save women money from having to repeat rounds after

unsuccessful IVF. However, although this technology is an exciting new frontier for reproductive

healthcare, there are still ethical challenges related to questions of who exactly gets access to this

technology as well as skepticism on how the samples to train AI models incorporate diversity which

continue to be explored.

References:

1. Koplin, J., Johnston, M., Webb, A. N. S., Mills, C., & Whittaker, A. (2025, January 7). Ethical

concerns over AI choosing which embryos may be born. Monash University.

https://www.monash.edu/news/articles/ethical-worries-when-using-ai-to-choose-embryos-in-assis

ted-reproduction

2. Chow, D. J. X., Wijesinghe, P., Dholakia, K., & Dunning, K. R. (2021). Does artificial

intelligence have a role in the IVF clinic?. Reproduction & fertility, 2(3), C29–C34.

https://doi.org/10.1530/RAF-21-0043

3. Olawade, D. B., Teke, J., Adeleye, K. K., Weerasinghe, K., Maidoki, M., & David-Olawade, A.

C. (2025). Artificial intelligence in in-vitro fertilization (IVF): A new era of precision and

personalization in fertility treatments. Journal of Gynecology Obstetrics and Human

Reproduction, 54(3), Article 102903. https://doi.org/10.1016/j.jogoh.2024.102903

4. Yale Medicine. (n.d.). In vitro fertilization (IVF). Yale Medicine. Retrieved April 14, 2025, from

https://www.yalemedicine.org/conditions/ivf

5. Murria, L., Bori Arnal, L., Sánchez Chiva, E., Albert, C., Cobo, A., & Meseguer, M. (2023).

Artificial intelligence oocyte image analysis predicts fertilization, blastocyst development, and

live birth outcomes per cohort. Fertility and Sterility, 120(4), e42–e43.

https://doi.org/10.1016/j.fertnstert.2023.08.151

6. Rajendran, S., Brendel, M., Barnes, J., Zhan, Q., Malmsten, J. E., Zisimopoulos, P., Sigaras, A.,

Ofori-Atta, K., Meseguer, M., Miller, K. A., Hoffman, D., Rosenwaks, Z., Elemento, O.,

Zaninovic, N., & Hajirasouliha, I. (2024). Automatic ploidy prediction and quality assessment of

human blastocysts using time-lapse imaging. Nature Communications, 15(1), Article 7756.

https://doi.org/10.1038/s41467-024-51823-7​:contentReference[oaicite:0]{index=0}

7. AIVF. (n.d.). EMA platform. Retrieved April 14, 2025, from https://aivf.co/ema-platform/

8. Merkison, J. M., et al. (2023). Racial and ethnic disparities in assisted reproductive technology: A

systematic review. Fertility and Sterility, 119(3), 341–347.

https://doi.org/10.1016/j.fertnstert.2023.01.023

9. Bierer, B. E., Meloney, L. G., Ahmed, H. R., & White, S. A. (2022). Advancing the inclusion of

underrepresented women in clinical research. Cell reports. Medicine, 3(4), 100553.

https://doi.org/10.1016/j.xcrm.2022.100553

Comment