'Bionanotechnology from Theory to Practice' is a short online, course providing an interdisciplinary and up-to-date overview of the rapidly developing area of bionanotechnology
Page URL: https://www.bionews.org.uk/page_143956

Benefits and challenges of using artificial intelligence in assisted reproductive technologies

22 July 2019
Appeared in BioNews 1007

First coined by a group of researchers in 1955, put simply: Artificial Intelligence (AI) is the science of making machines do things that would require intelligence if done by humans. Since then, AI has developed rapidly and its applications have permeated our daily lives including automated vehicles, facial recognition and intelligent voice assistance. AI is also used in healthcare, with fields such as oncology, radiology and cardiology benefiting from its applications. 

AI made its debut into the research world of assisted reproductive technologies (ART) in the late 1990s, through the creation of an algorithm aimed at predicting the outcome of IVF. More AI technologies using different kinds of algorithms followed and were used in various ways, including sperm cell classification and oocyte and embryo selection. 

In a recent study, researchers trained an AI algorithm to identify, with 97 percent accuracy, high and low-quality embryos. The AI algorithm outperformed, in an objective manner, the individual embryologists responsible for assessing embryo quality using morphological analysis - a subjective evaluation involving manual grading of the human embryo. Although the algorithm cannot predict pregnancy rates, the accurate information it can provide about embryo quality is a crucial variable among others (such as age) that may improve a couple's chance of conceiving.

The potential introduction of AI into the clinical ART world holds both tremendous benefits and ethical complexities. If approved for clinical application, the use of AI to separate high-quality embryos from those that are chromosomally abnormal might save healthcare professionals time and effort by processing and interpreting more data with greater depth and precision. This might, in turn, improve the efficiency of ART and subsequent pregnancy outcomes, treatment options and care for patients with infertility. At a societal level, it could minimise healthcare costs by reducing the use of unnecessary testing or treatment. 

As the use of AI in ART moves rapidly from research to the clinic, all stakeholders, including the general public, decision-makers, clinicians, and scientists, should anticipate and reflect upon the potential ethical challenges that might arise. Some of these challenges are common to those raised by the application of AI in other healthcare domains. 

For instance, it is feared that AI techniques might replace health professionals if their performance exceeds that of an expert clinician. We should be aware that AI is as an auxiliary tool to assist health professionals in offering high-quality healthcare in a more efficient and accurate manner. However, it does highlight a need to better educate future health professionals by integrating AI technology into the curriculum. 

Other considerations are more specific to ART. For example: what if embryo selection performed by the AI algorithm went wrong and thus the embryo chosen appeared to be 'abnormal' following its implantation or during pregnancy. Who would be held responsible for this unexpected output: the health professional or the AI? And what measures (legal, organisational, ethical, etc.) should be implemented in order to avoid such situations or/and resolve them? 

If AI is currently trained to select the good-quality or 'normal' embryos, could it be potentially trained and used to 'select' embryos with the 'best genetic make-up'? Or could it be used to 'design' embryos with the 'best genes'? 

For now, the introduction of AI techniques is clearly revolutionising healthcare and could transform both the experience of couples as well as medical practice. However, there are still many ethical challenges to be considered and overcome.

Some guidelines, policies and recommendations are already in place to offer an ethical framework that can guide the use and implementation of AI technologies in the clinic. However, the sweeping speed at which AI techniques are being developed or sometimes used, even before the publication of appropriate policies and guidelines, might leave users (such as clinicians) confused about how to best integrate this new technology into their practice. Consequently, developing up-to-date guidelines as well as education will be crucial as AI techniques become more widely available and complex.

Artificial intelligence and machine learning for human reproduction and embryology presented at ASRM and ESHRE 2018.
Journal of Assisted Reproduction Genetics |  28 January 2019
Deep learning enables robust assessment and selection of human blastocysts after in vitro fertilization
NPJ Digital Medicine |  4 April 2019
Get on the AI and big data bandwagon!
Impact Ethics |  2 April 2019
The application of neural networks in predicting the outcome of in-vitro fertilization.
Human Reproduction |  1 July 1997
6 April 2020 - by Dr Charlott Repschlager 
A new blood test has been found to detect over 50 types of cancer, using a machine learning algorithm system and circulating tumour DNA...
17 June 2019 - by Shaoni Bhattacharya 
GlaxoSmithKline is to partner with leading researchers at the University of California to harness CRISPR/Cas9 genome-editing to find new medicines...
8 April 2019 - by Bethany Muller 
Researchers have trained artificial intelligence software to grade human embryos with a high degree of accuracy...
22 October 2018 - by Tara Schmidlen and Amy Curry Sturm 
It's 2018, and Alexa has become a common household name. The iPhone just turned 10 and roughly two-thirds of adults worldwide own a smartphone. About 84 percent of the global population lives in an area where mobile broadband access is available. At the same time, genetic testing has become more affordable and accessible to consumers, with direct-to-consumer genetic testing pioneer, 23andMe, now boasting five million users in over 50 countries...
6 November 2017 - by Isobel Steer 
Machine learning has been used to develop a promising screening test for ovarian cancer, a team of US-based researchers has said...
to add a Comment.

By posting a comment you agree to abide by the BioNews terms and conditions

Syndicate this story - click here to enquire about using this story.