Digital transformation of human reproduction

Consider This
Digital transformation of human reproduction


Neil Kenneth McBride

De Montfort University

Consider This:

As digital transformation permeates working and personal life, the intention is that no aspect of human life is untouched. For a significant number of couples fertility is an issue. Down the line this can result in referral for in-vitro fertilisation (IVF) treatment. In IVF, following suppression of the menstrual cycle, and follicle stimulation by hormone treatment, eggs are harvested and mixed with sperm to fertilise them.  This could involve injecting individual eggs with sperm. Embryos are grown in the laboratory and monitored daily. The best one or two are selected for transfer and implanted in the prepared womb using a catheter.[1] Usual practice is an elective single embryo transfer (eSET). [2]

How can such a process be digitally transformed? A possibility exists in the selection of the best embryo. On average 15 eggs are collected and the resulting embryo graded using a range of parameters. Partly this involves observation and judgement of the development of embryos. Time-lapse photography enables careful monitoring of changes embryos. Grading is a team effort. Early on at day three it simply involves rate of growth and degree of fragmentation. As the blastocyst develops cavity, central cell mass and the progression of the area that attaches to the placenta is monitored and graded.[3] It is not an exact science. It depends on the teams’ knowledge and practice within the clinic and built-up tacit knowledge.

Digital transformation requires bringing leading-edge technologies to bear on assisted reproductive technology. However, can we do this in an appropriate manner for the application and its context? Or are we really arguing for digital transformation without the justification of its value?  Since IVF requires the grading of embryos based on visual inspection or image analysis, is this not a task which could be carried out by AI? On face value, yes. If was have enough images of high-quality embryos we could train a neural net to identify good embryos. Zaninovic et al (2019)[4] trained a neural net on 50,000 time-lapse microscopy images which could identify quality embryos. The possibility is that AI can do a better job than embryologists.  AI may see details the embryologists don’t and eliminate operator subjectivity. It performs tedious repetitive task, without emotional or physical constraints. The authors claim that AI will bring digital transformation and automatization to the field of reproductive medicine. In a subsequent paper Rosenwaks is evens more effusive about the power of AI.  AI approaches will be more objective, more accurate and more rapid.  they will result in precision, standardisation, and automatization. Treatments will be done with precision and these treatments will individualise in ways which will be superior to what s possible by humans alone. Decisions will be objective, reproducible, repeatable compared to emotional and fatigued humans.

The digital transformation intended by the use of AI is intended to remove errors, to provide treatment which is superior to what human can do, But the point of the proposed digital transformation is to quantitate IVF. Is such quantitating digital transformation? What is being transformed? Is it appropriate?

A closer examination of IVF and embryo selection reveal complexities. It seems also that selecting embryos is both complex and variable according to clinics. Embryo selection is affected by other hidden factors. Hence generation from image analysis is not without its issues. Furthermore, the best embryo might not implant while a low-grade embryo results in a beautiful baby.[5] The difficult decision for the embryologist is whether to select an embryo that is well developed but doesn’t have a perfect grade, or to select a slightly less well-developed embryo that has an AA grade[6].The choice is a matter of discussion and the application of deep experience developed within a practice.

There is an almost religious focus on AI as something whose accuracy and power are such that human judgement is eliminated. In a sense this brings us to an epistemological ethics, a need for transparency and sober minded consideration of the systems which are IVF, and the complex human, technical and scientific interactions. What we try and do in AI is reduce complex human processes to sets of manipulatable data, which is all AI can deal with. This reductionism, in defining data sets, leaves much behind, and is rather dehumanising. The rhetoric is one of certainty, the AI is right because the data is right. The embryologist’s experience is gained in practice and discussion, in considering the complex systems surrounding the patient and developing tacit knowledge. Digital transformation by AI may deny the importance of the human involvement. Perhaps the use of AI here creates a false sense of certainty, of security.

AI requires large data sets, data set which are consistent and accurate. It requires something easily generalisable. In the case of IVF, a maximum of 20 eggs are harvested and the growing of embryos is almost a personal relationship as the embryologist monitors the embryos, makes informed judgements and discusses within a multidisciplinary team, incorporating the complexities of counselling, genetics and medical issues. And if we’re only examining a small set of embryos, perhaps complex image analysis by a neural network might not offer many new insights. Clinics differ, procedures vary and calibration between clinics and even time lapse photography processes may differ in small ways.  

IVF is of its nature, relational, emotional and ethically charged. Parents will form a close relationship with a clinic where the role of the expert, located in the team is key. They make the decision to trust the clinic which is assisting the parents in their goal., The process is relationally embedded in the culture of the parents and the society. A focus on automatization, cost and efficiency misses the point.

The study of the use of AI in IVF raises the question not only of what constitutes digital transformation, but where digital transformation, if that is what the use of AI constitutes, is relevant. Assessing whether AI is the right solution requires not just an analysis of the scope and nature of the data, but also a careful consideration of the nature of the application[7] We cannot take it that any application of leading-edge technology constitutes digital transformation, nor that all business and processes are simply inadequate without digital transformation.

Additionally, rhetoric around digital transformation which emphasises the accuracy and dependability of the technology and downgrades the role and importance of human intervention should be questioned. Image interpretation is only as good as the data and the interpretation provided by experts. And it is the exceptions which cause the problems and require the tacit knowledge and experience of the expert.

If we are dealing with thousands of retinal scans and wish to eliminate straightforward cataracts to leave the exception for the ophthalmologists, a digital transformation makes sense. Monitoring the growth of three embryos over six days is an entirely different matter. It is not only definitions of digital transformation which should be evaluated, but processes which are being subjected to digital transformation.


[1] NHS (2022) IVF: What Happens

[2] Human Fertilisation and Embryology Authority (2022) Decisions to make about your embryos

[3] ORM Fertility (2019) Embryo Grading: The Good , the poor and the baby making kind.

[4] Zaninovic N., Elemento O., Rosenwaks Z.: Artificial intelligence: its applications in reproductive medicine and the assisted reproductive technologies. Fertil Steril. 112(1), 28-30 (2019). doi: 10.1016/j.fertnstert.2019.05.019.

[5] Advanced Fertility Centre of Chicago (2022) IVF Embryo Quality and day 3 embryo grading after in vitro fertilisation: Cleavage stage embryo grading.

[6] Bourn Hall (2018) How do you choose an embryo for IVF?

[7] Office of Artificial Intelligence (2019) A Guide to Using Artificial Intelligence in the Public Sector

Please sign in or register for FREE

Your Fertility and Sterility Dialog login information is not the same as your ASRM or EES credentials. Users must create a separate account to comment or interact on the Dialog.