With in vitro fertilisation (IVF) success rates stubbornly hovering around the 30 to 40 percent mark for decades, much focus has been given to embryo selection and testing. In recent years, developments in this area have raised ethical and legal concerns, including the use of artificial intelligence (AI) and the adoption of new methods of genetic testing by Monash IVF.
In the third insight of our ART in Focus series, we consider how the area of embryo selection and testing engages ethical and regulatory considerations, and explore what this means for assisted reproductive technology (ART) providers moving forward.
Embryo selection and testing in Australia
Embryo selection and testing in Australia is a growing area of ART which aims to improve the success rate of IVF, a popular treatment for infertility.
The process of IVF involves using medication to trigger ovulation, following which eggs are extracted from the ovaries and fertilised by sperm in a laboratory, to create a human embryo. The embryo is subsequently transferred into a uterus with a view to progressing to pregnancy and ultimately a live birth.
Embryo selection and testing is conducted following the fertilisation and prior to transfer, and involves the review of embryos to assess quality and determine which embryo/s should be selected for transfer.
However, this is not without controversy. In Australia, pre-implantation genetic testing is restricted to limited purposes, including testing for diseases that would severely limit the quality of life for the person born.1 In contrast, testing in the United States at this stage can be used to screen for genetic predispositions to common diseases, including bipolar disorder, diabetes and coronary artery disease, which are recognisable through screening the embryo genome.2
In recent times, embryo selection technology has developed to incorporate machine learning (a type of AI) that analyses ‘time-lapse images and/or videos of embryos at various stages of development, and then providing predictions of embryo quality based on the embryo’s morphology and morphokinetics’.3 In conjunction with the use of AI, embryologists (the scientists responsible for fertility treatment) grade each embryo.
Some early studies have suggested that machine learning AI may outperform human embryologists, with the potential to increasingly subsume the role for human intervention.4 For example, a literature review published in 2023 reported that machine learning AI consistently outperformed clinical teams in studies on embryo morphology and clinical outcome prediction during embryo selection assessments.5
Machines choosing human life? Ethical and legal concerns
Some observers have raised ethical concerns regarding the use of machine learning AI in the process of embryo selection and testing, including researchers from Monash University which published a paper in 2024 exploring these issues.6
Informed consent and explainability
The concept of informed consent is fundamental to the lawfulness of medical treatment, and concerns have been raised about how to ensure informed consent within the context of using AI for embryo selection.
Informed consent requires that consent is given by a person with capacity, it must be freely and voluntarily given, and it must cover the act to be performed.
However, critics have suggested that black box machine learning7 is not compatible with the values and principles of informed consent, given the lack of explainability in the process.8 Where AI is used within the context of medical procedures, the ability to understand the act to be performed is challenging, as the internal reasoning of an algorithm is not always easily understood.
As the use of AI in embryo selection processes increases, so do the calls for steps to ensure transparency for consumers of IVF as a prerequisite for informed decision-making.9 Future safeguards could include industry-specific guidelines for informed consent in the use of AI, which set out minimum information requirements for consumers about the use of AI in the embryo selection process.
Deskilling embryologists
Concerns have also been raised regarding ‘scope creep’ by AI, and the potential for deskilling of clinicians, specifically, embryologists, over time. As with other areas of healthcare - as evidence mounts to suggest technology can outperform human expertise, challenges arise as to how to ensure, in the absence of specific regulation, that there remains a ‘human in the loop’.
Related to this is the potential for embryologists in certain clinics to become less skilled in certain tasks over time, where such tasks become reliant on the use of AI. The need for human embryologists and the upkeep of skills is not only an important safeguard during the early stages of AI integration, but is crucial to ensure that skills are maintained within the profession in circumstances where embryologists may move to different workplaces where they are required to assess embryos manually.
This has been identified as a particular concern given AI cannot be used worldwide, as not all clinics have access to the necessary resources.10
Presently, embryologists in Australia are not regulated health professionals, unlike in New Zealand or the UK. Accordingly, they are not subject to formal continuing professional development obligations which require them to maintain skills and education.
Who is responsible?
Another concern in relation to using AI for embryo selection and testing lies in identifying the responsible party in the event of an adverse outcome. A feature of AI tools is that they largely ‘teach themselves’ and are not controlled by the manufacturer or programmer.11
This is a common concern within many areas of medicine, but has attracted particular scrutiny in the areas of embryo selection and testing, which involve decisions surrounding the creation of life. A valid question can be raised regarding who should be held responsible if the implementation of AI embryo selection and testing either ‘increases average time to pregnancy or reduces live birth rates (e.g. if viable embryos are discarded)’.12
Other concerns
Other concerns raised include the potential for the process to ‘dehumanise’ embryos, and that perceptions of a diminishing role for human oversight or intervention have the potential to erode patient trust and decrease confidence in the decision-making and critical thinking behind the procedures.
Innovation in embryo testing and the Monash class action
As outlined above, considerable focus has been directed in recent years to developing new methods to improve embryo selection and test for embryo viability in order to improve IVF success rates. While it is an area ripe for innovation, the implementation of new technologies carries risks for providers and developers. Even in cases where the technology itself is regarded as ‘non-invasive’, the consequences of inaccurate testing may be significant given the potential for destruction of viable embryos and reduction of successful outcomes.
The Monash class action settlement in August 2024 for $56 million is one example of this. Starting in 2020, more than 700 patients were involved in a class action against Monash IVF that involved the service of cell-free non-invasive pre-implantation genetic testing (NiPGT-A testing) of live embryos instead of biopsy testing to determine the aneuploid status of embryos. The testing was promoted as superior to biopsy testing because the non invasive nature minimised the risk of causing damage to the embryos.
The claim asserted however that the testing should not have been utilised as there was an increased risk of false positives and false negatives not disclosed to patients prior to the service.13 In addition, the patients alleged that Monash IVF had used the embryos that they had instructed the patients to discard without appropriate consent.14
Monash IVF settled the case with no admission of liability, however it was later admitted in a revised fact sheet that the now-suspended NiPGT-A testing was giving the standard test results only 75 to 85% of the time.15
This case highlighted the need for integrity and robust peer review in research prior to implementing a new technology, and the scope for class actions to arise as a consequence of new technology in the ART space.
Key takeaways
As the above discussion outlines, embryo selection and testing is an area which raises several legal and ethical concerns, particularly as new technologies are introduced.
Both expert and public opinion emphasise the need for considered decisions to be made to maintain good clinical practice and determine accountability for clinical outcomes, highlighting the need for continuous monitoring of AI.
Regulation and legislative change are often touted as the answer to emerging risks, however as outlined in the second article of this series ART in Focus: The donor conception minefield, the legal and regulatory environment in this area is the subject of review which may take some time. Meanwhile, providers should be carefully considering their processes for informed consent and ensuring their use of AI aligns with existing guidelines for use of AI.16
Providing clear and transparent information to consumers regarding proposed new technologies and their risk/benefit profile upfront will reduce the risk of claims and complaints. Providers also need adequate protocols at their clinics to ensure that there remains a ‘human in the loop’ when AI is implemented, and that their embryologists maintain their key skills and competence.
While innovation in embryo selection and testing is crucial to continued improvement, care must be taken when adopting new technologies for embryo selection given the potential for widespread impacts, as illustrated by the Monash class action. The literature supporting new developments should be interrogated and providers ought to satisfy themselves of the integrity of research methods.
Finally, ethical concerns remain a key aspect of public confidence in the industry. Partnering with consumers by transparent disclosures and effective consenting processes remains a key part of implementing any innovation in healthcare.
1 Ethical guidelines on the use of assisted reproductive technology in clinical practice and research 2017 (updated 2023)
2 Designer Babies? The Ethical and Regulatory Implications of Polygenic Embryo Screening
3 Ethics of artificial intelligence in embryo assessment: mapping the terrain
4 Ethical concerns over AI choosing which embryos may be born - Monash University
5 Embryo Selection through Artificial Intelligence Versus Embryologists: A Systemic Review
6 Ethics of artificial intelligence in embryo assessment: mapping the terrain
7 When machines malpractice: Liability for AI and robotic healthcare in Australia
8 Ethics of artificial intelligence in embryo assessment: mapping the terrain
9 Embro Selection, AI and reproductive choice
10 Ethics of artificial intelligence in embryo assessment: mapping the terrain
11 Ethics of artificial intelligence in embryo assessment: mapping the terrain
12 Ethics of artificial intelligence in embryo assessment: mapping the terrain
13 Monash IVF Group Proceeding Summary Statement filed 4 March 2021.pdf
14 Fertility giant Monash IVF settles class action with more than 700 patients over inaccurate embryo screening test - ABC News
15 Monash IVF Group's 'inaccurate' genetic test potentially robbed these women of viable embryos - ABC News
16 When machines malpractice: Liability for AI and robotic healthcare in Australia
