Embryo Selection in the Context of In Vitro Fertilization

Abstract

From the Catholic perspective, in vitro fertilization (IVF) is morally problematic because it artificially separates the procreative and unitive aspects of the conjugal act. Embryo selection (ES) in the context of IVF is an injustice against the resulting embryos because it treats them as commodities and works against their right to life by determining their implantation potential in light of their features. The Church opposes the eugenics mentality underlying ES. Meanwhile, the IVF industry increasingly uses artificial intelligence (AI) for ES. However, doing so could worsen the injustice by deepening the disrespect of human lives under the technocratic paradigm. As such, Catholic bioethicists are encouraged to advocate for the Church's teachings with renewed vigor. In this commentary, we will examine (1) ES in the context of IVF, (2) using AI for ES, (3) the moral implications of using AI for ES, and (4) points for further consideration.

Summary: Using AI for Embryo selection in the context of IVF deepens the disrespect of human lives under the technocratic paradigm.

Infertility is a medical condition that is defined by the United States Centers for Disease Control and Prevention “as not being able to get pregnant (conceive) after one year (or longer) of unprotected sex” (n.d.). Scientists developed medical techniques collectively known as assisted reproductive technology (ART) to bypass the physical limitations of infertility to achieve pregnancy. The most popular ART type is in vitro fertilization (IVF). Generally, IVF begins with preliminary tests conducted on both the woman and man to evaluate the health of their respective ovaries, eggs, uterus, and semen. Afterward, several general stages follow (Collington 2007). The first is ovulation induction or hyperstimulation of ovaries through hormones to induce the production of eggs more than usual. Their retrieval follows this using an ultrasound-guided needle entering the vaginal cavity. Around the same time, semen is collected from the husband, partner, or anonymous vendor, often by masturbation. The eggs and sperm cells are then fertilized in nutritive material inside a petri dish in a ratio of 1:50,000–100,000, respectively. As a result, multiple zygotes form. They are then left to develop in culture, and time-lapsed photographs are taken. In about five days at the blastocyst stage, an embryologist evaluates the photographs and screens the embryos using a high-powered light microscope to check their morphology and developmental rate in culture for abnormal developments to determine their implantation potential based on a grading system. The selection criteria normally “include the number of blastomeres, the absence of multinucleation, early cleavage to the two-cell stage, and a low percentage of cell fragments in embryos” (Wang 2011, 488). The rationale of embryo selection (ES) is that there would be better chances of a successful pregnancy and live birth if abnormal embryos were excluded. The embryos with a high implantation potential can also undergo a biopsy for genetic testing to identify the presence of underlying diseases. One to three embryos are implanted in the woman's uterus for an increased likelihood of pregnancy. Any remaining ones are cryopreserved for future use in case the initially implanted embryos fail to gestate or the woman desires another child. If more embryos gestate than is desired, the woman can request a pregnancy reduction which eliminates excess embryos. A single IVF cycle can take about three months with an expense of thousands of dollars. The woman can undergo two to three IVF cycles to improve her chances of pregnancy. Despite the measures taken, the current success rate of IVF pregnancy is below 50% and that of live birth is around 30%.

Using Artificial Intelligence for ES

Artificial intelligence (AI) is “the field of study devoted to making machines intelligent. Intelligence measures a system's ability to determine the best course of action to achieve its goals in a wide range of environments” (Scharre and Horowitz 2018). In other words, it is “the science and engineering of making intelligent machines, especially intelligent computer programs” (McCarthy 2011). A sub-branch of AI is known as connectionist (or emergentist) AI. It specializes in analyzing data and recognizing desired patterns, such as when researchers input various images of an object and train the AI to associate them with a particular output. The AI system soon “learns” to act alone according to the patterns with greater precision achieved when more data is available. This independent learning process is known as Machine Learning. It does this by “systematically [applying] algorithms to synthesize the underlying relationships among data and information” (Awad and Khanna 2015). A subset of Machine Learning is Deep Learning, which “is a special kind of learning with deep artificial neural networks” (Skansi 2018, v) inspired by the human brain. An artificial neural network “is a set of simple processing elements connected together to form a network of nodes that uses a mathematical model for information processing” (Manna et al. 2013, 44). Today, Deep Learning “has become established as one of the most impactful research areas within [AI]” (Garnelo and Shanahan 2019, 17). However, some issues of connectionist AI are that it takes a long time to process, it does not provide a clear step-by-step layout of the learning process (black box problem), and it may produce poor or biased output if the datasets are deficient.

Connectionist AI is helpful in many industries, especially in healthcare. Fidelma Fitzpatrick et al. state that “AI has the potential to detect transmission events during outbreaks or predict high-risk patients, enabling [the] development of tailored [infection prevention and controlled] interventions” (2020, 1). They give the example of Bluedot, an AI platform, which “first alerted on COVID-19 on December 31, 2019, almost a week ahead of national surveillance centers and WHO” (2020, 4). Eventually, more advanced AI could help prevent pandemics from occurring. Furthermore, Anita Ho suggests that AI could assist in elderly care. The global population of those aged at least sixty-five years may reach 1.6 billion by 2050, with no guarantee of a commensurate increase in healthcare professionals (Ho 2020, 1). Regarding this, “AI monitoring data regarding older adults’ risk levels for adverse events can inform medical decision making and transform healthcare delivery. They can help healthcare providers to triage cases to ensure that the right patients have timely access to appropriate care” (2020, 2).

With the popularity of ART in the healthcare industry, it is unsurprising that AI is playing an increasingly significant role in it. Curoche and Bormann observe that “In just 1 year [from 2017 to 2018], published abstracts [on AI and Machine Learning] at the annual proceedings of the American Society of Reproduction and European Society of Human Reproduction and Embryology increased seven-fold [from two to sixteen]” (2019, 591). In 2022, the European Society of Human Reproduction and Embryology had at least twenty papers about AI and ART. In the context of IVF, AI can assess the health of sperm and oocyte, help predict pregnancy success, and help “automatic annotation of embryo development (cell stages and cell cycles), embryo grading (mostly in the BL stage), and embryo selection for implantation” (Zaninovic and Rosenwaks 2020, 915). ES is an ideal use for Machine Learning and Deep Learning because it involves examining thousands of time-lapsed microscopy embryo images to determine output, namely embryo quality. The output can be used as a reference to classify the suitability of embryos for genetic testing and eventual implantation. In doing so, more data is provided to refine the accuracy of the AI system. A shared data system of embryo images could eventually be made publicly available for different researchers to use in their AI systems and algorithms. Thus far, research suggested a slightly better correlation between predicted embryo quality and implantation potential if done by AI than by embryologists (Kragh et al. 2019). Another research retrospectively used data on predicted embryo quality and live birth to develop an efficient AI model without manual embryo annotation and evaluation (Huang et al. 2022). One other research reported on the efficient validation of an AI model using time-lapsed images of embryos with known ploidy status for eventual triage use in preimplantation genetic testing (Meseguer Escriva et al. 2022). But one research cautioned that the predictive power of AI for pregnancy outcome and/or selecting the embryo with the highest implantation potential is still not ready for a routine IVF laboratory setting (Simopoulou et al. 2018, 1554).

Moral Implications of Using AI for ES

Since ES occurs in the context of IVF, a summary of the teachings of the Church regarding the latter is helpful. The Church considers IVF unacceptable because it does not assist the conjugal act but replaces it entirely. While it is not the focus of this commentary to elaborate on this teaching, suffice it to say that the Church teaches that a child's conception can morally occur only within the conjugal act (sexual intercourse in marriage). The conjugal act has two essential aspects, which are procreative (openness to the transmission of life) and unitive (physical self-giving of the spouses to each other). The Church cautions that “[n]ever is it permitted to separate these different aspects to the point of excluding positively either the intention of procreation or the conjugal relation” (Pius XII 1956). To artificially separate them is to intentionally inhibit the full expression of personal love through the body's language. With the IVF industry continuing momentum, the Church must better persuade the public on respecting the two essential aspects of the conjugal act. Once the IVF occurs, the focus turns to the IVF embryos. Their wellbeing is of urgent importance since they are vulnerable to health issues outside their natural environment, which is in their mother. From the Catholic perspective, human beings have dignity from the moment of their fertilization. This moment is not an arbitrary one but is scientifically grounded. Informed by the medical sciences, the Church teaches that “from the time that the ovum is fertilized, a new life is begun which is neither that of the father nor of the mother; it is rather the life of a new human being with his [or her] own growth” (CDF 1987). It adds that “in the zygote resulting from fertilization the biological identity of a new human individual is already formed” (CDF 1987). As such, the Church strongly asserts that:

The human being is to be respected and treated as a person from the moment of conception; and therefore, from that same moment his [or her] rights as a person must be recognized, among which in the first place is the inviolable right of every innocent human being to life. (CDF 1987)

The embryo's human dignity must be central in morally analyzing AI for ES. For clarity, not all prenatal assessments are morally problematic. It is morally acceptable if it “does not threaten the life or physical integrity of the unborn child or the mother and does not subject them to disproportionate risk” (CDF 1987). In other words, prenatal assessment is morally acceptable if safely done in vivo strictly for therapeutic purposes. Using AI as part of a therapeutic prenatal assessment on an embryo in vivo may be admirable. But there must also be a clear intention to continue the pregnancy if the results reveal a high likelihood of abnormal development. Unfortunately, such is not the case for ES. As noted, IVF embryos undergo ES before implantation to ensure optimal results for pregnancy, and those with a high likelihood of abnormal developments are likely to be medically discarded. Thus, ES is an injustice because it works against the right to life of the embryos by determining their implantation potential in light of their features. The Church opposes the eugenics mentality underlying ES. Pope John Paul II remarks in his encyclical Evangelium Vitae that this eugenics mentality:

[A]ccepts selective abortion in order to prevent the birth of children affected by various types of anomalies. Such an attitude is shameful and utterly reprehensible, since it presumes to measure the value of a human life only within the parameters of ‘normality’ and physical well-being, thus opening the way to legitimizing infanticide and euthanasia as well. (1995, 63)

Understandably, the “selective abortion” referred to often occurs ex utero in the context of IVF since the embryos were fertilized outside of their natural environment. Even if the woman expresses the clear intention to keep them should the results reveal a high likelihood for abnormal development, ES would still be morally problematic because, per se, it objectifies the embryos by treating them as commodities subject to quality control. Even if one were to assert that IVF is not morally problematic, the same assertion does not hold with ES because it intentionally undermines their right to life. There is little point in ES if the medical presumption is that all embryos have the right to an equal opportunity at implantation. Thus, supposing one disagrees with the Church on the ethics of IVF, there is still reason to disagree morally with ES. Unfortunately, there is a lack of appreciation regarding the Catholic perspective on these matters. As a result, ES routinely occurs at the expense of the human dignity of the embryos. Finally, it might be argued that to prevent ES entirely, there must be health assessments as early as the germ cell stage to prevent the fertilization of abnormally developed embryos, but the rationale for doing so would still give the impression that embryos are commodities subject to quality control.

Concerning the use of AI, the means employed and the circumstances involved can aggravate morally problematic acts such as ES. There is a moral difference between unjustly attacking in front of someone with bare fists and unjustly attacking from behind with a baseball bat, the latter being treacherous. There is also a moral difference between unjust theft from a private person and unjust theft from a church, the latter being sacrilege. In this light, using AI could worsen the injustice of ES by deepening the disrespect of human lives under the “technocratic paradigm” cautioned against by Pope Francis in his encyclical Laudato Si’. There, he observes that:

[T]he way that humanity has taken up technology and its development according to an undifferentiated and one-dimensional paradigm exalts the concept of a subject who, using logical and rational procedures, progressively approaches and gains control over an external object. (2015, 106)

Although the term “technocratic paradigm” in Catholic social teachings was popularized by Laudato Si’, the concept underlying it, arguably, had begun much earlier. The steady progression of industrialization and globalization over the past two to three hundred years has brought human achievement to new levels. The technological wonders that humanity has created in that period have made life easier and more convenient, even though the distribution of these technologies to less privileged parts of the world is still a work in progress. Using technology is not necessarily technocratic. The Church teaches that science and technology should be for the common good of the human family (Catholic Church 1992, 2293). Created in God's image and likeness, humanity collaborates in realizing God's creative vision through science and technology. Today, technology is capable of interpreting vast quantities of information with relative speed, but this epistemic simplification risks eroding our capacity for deep ethical reflections. Excessive reliance on technology leads to a technocratic attitude that shortens the gap between what can and what must be done, with little thought for the ethical implications.

This technocratic attitude has affected even the healthcare industry. According to Travaline, “There are two major overarching moral dangers with respect to technocratic medicine. One is a flawed anthropology that causes and perpetuates a crisis of the human person and the second concerns the medicalization of death” (2019, 232). The flawed anthropology treats the human body as an instrumentalized machine. As a consequence, health is viewed mainly in terms of statistics. On the other hand, the medicalization of death views death as a failure rather than a natural part of the human experience, albeit an often painful one. In both senses, IVF is an example of the technocratic paradigm. The limits of the human body are being toppled one by one through technology. We view sexuality not as a gift from God but only as a complex challenge to dominate. The technocratic paradigm promotes the belief that technology can and must overcome infertility, even if it involves fertilizing and medically discarding thousands of embryos. The deaths of these innocent lives within the IVF industry have just become more problems to resolve through the use of even more technology, that is, AI. Ironically, advancements in technology have revealed more and more about the complexity of embryos, which should persuade us more about their humanity. Instead, we risk further losing sight of it through a misguided technological imperative. The irony of using AI for ES is pronounced given the joint statement signed in 2020 by the Pontifical Academy for Life and several tech companies entitled “Rome Call for AI Ethics” (Pontifical Academy for Life 2020), which explicitly states that:

All human beings are born free and equal in dignity and rights. They are endowed with reason and conscience and should act towards one another in a spirit of fellowship (cf. Art. 1, Univ. Dec. Human Rights). This fundamental condition of freedom and dignity must also be protected and guaranteed when producing and using AI systems. This must be done by safeguarding the rights and the freedom of individuals so that they are not discriminated against by algorithms due to their “race, color, sex, language, religion, political or other opinion, national or social origin, property, birth or other status.” (Art. 2, Univ. Dec. Human Rights)

An AI system used in ES can conclude that a particular embryo has a high likelihood of abnormal development, but it cannot adequately explain why. While an embryologist can verify the result, the presumption is already there. The fate of an embryo fertilized through technology is subject to the decision of technology. A reasonable response to this dilemma would be not to use AI for ES, yet the technocratic paradigm has proven difficult to resist. The more we do technically, the less likely we desire to turn back out of convenience. Digital calculators are being used today for basic arithmetic computations. If anything, using AI gives the impression of technical objectivity in ES with no human subjectivity involved and is, therefore, an improvement for IVF. Supporters may argue that AI could prevent unnecessary embryo wastage by being more accurate than an embryologist, who could misdiagnose some embryos as false negatives. Healthcare facilities would then be emboldened to outdo each other by adopting more AI in their IVF services. But the truth is that AI raises issues, including data privacy, yet to be properly addressed by scientists and ethicists, regardless of effectiveness in preliminary research. The claim of objectivity is put into question by the fact that there is no single AI system and algorithm consistently being used among healthcare facilities worldwide. It does not, however, mean that AI would be acceptable for ES after the technical issues are addressed. If an act is morally problematic, as in the case of ES, then it should be avoided. Still, if an embryologist or AI alone conducting ES is an injustice, could both provide a two-layered process that ensures more embryos have a chance at life? If so, the order of the two would be essential. As noted earlier, AI conducted first might negatively influence the judgment of an embryologist. On the other hand, AI could prevent false negatives determined by an embryologist. With the likelihood of increasing use of AI in ES, Catholic bioethicists must look for new ways to advocate for the interests of the embryos.

Points for Further Consideration

We need to consider four more points to arrive at a just evaluation of the issues presented thus far. (1) The Church supports responsible scientific and technological research and development. As noted, science and technology are means for us to collaborate in the creative vision of God. We must view Church critique or caution vis-à-vis morally problematic assisted reproductive technology in the context of the invitation of Jesus to Christian discipleship (Levada 2007). Rather than view the bioethics of the Church as primarily prohibitive, we should instead view it as a collaborative effort in promoting human flourishing while respecting our theological anthropology constituted as body and soul. Exactly fifty years before Laudato Si’, the pastoral constitution of the Second Vatican Council, Gaudium et Spes, cautioned against the technocratic paradigm by challenging the whole Church to:

Let [the faithful] blend new sciences and theories and the understanding of the most recent discoveries with Christian morality and the teaching of Christian doctrine, so that their religious culture and morality may keep pace with scientific knowledge and with the constantly progressing technology. Thus they will be able to interpret and evaluate all things in a truly Christian spirit. (Vatican II 1965, 62)

The Church welcomes scientific research which aims to treat infertility and associated medical conditions without undermining our theological anthropology. Moreover, the Church is generally open to the ethical use of AI technology in health services, especially when it can supplement the noble work of healthcare professionals by giving them more time to focus on more urgent medical matters. (2) The Church upholds the legitimate right to privacy for everyone. AI must not infringe on this right, even in issues of morally problematic ART and ES. Healthcare professionals must ensure that their research and conduct are ethically sound in this regard. (3) The Church encourages infertile couples to consider traditional adoption rather than IVF to fulfill their desire to be parents. The injustice of ES would not have occurred if the IVF did not happen. But for IVF embryos already in existence, their human dignity would, arguably, be best advanced if they were all promptly implanted in their mother's uterus, even with the possibility of abnormal development. (4) The Church understands the stress and anxiety associated with bringing to term embryos with a high likelihood of abnormal development. Still, it advocates for their right to life. As a matter of social justice, the Church encourages governmental and nongovernmental efforts to provide more accessible and affordable means to parents and guardians for raising children with special needs. Such children can live fulfilling lives, and many do.

In conclusion, AI is poised to play an increasingly significant role in ART, including ES, over time, so appropriate moral guidance must follow. While this commentary does not exhaust the issues presented, I hope it provides ideas for moral guidance and elicits interest in learning more. As things are, Catholic bioethicists already have difficulty persuading the public on the moral problems of IVF and other similar ART types. As the technocratic paradigm in ART deepens with using AI for ES, the responsibility of Catholic bioethicists to advocate for the Church's teachings will be all the more important even in the face of doubt and strong opposition. I encourage Catholic bioethicists to continue their advocacy with renewed vigor.