Professor of Bioethics Monash University Clayton, Victoria
Abstract: Machine learning (ML) systems are increasingly being integrated into assisted reproductive technology (ART) clinics, particularly for the assessment and selection of embryos prior to transfer to a gestator’s uterus. While a growing literature addresses the ethical and social implications of machine learning for embryo assessment (MLEA)—especially regarding transparency, patient disclosure, and consent—little is known about the perspectives of ART professionals who use or will soon use these systems.
This paper presents findings from an anonymous online survey of clinical scientists working in Australian ART clinics conducted in September-October 2023. We focus on findings relevant to transparency and professional and patient trust in MLEA technology.
Among the 51 survey participants, results show that scientists highly value model transparency, with nearly all respondents stating that understanding the factors informing the model's assessment is essential to their professional roles and trust in the technology. However, over half indicated willingness to sacrifice transparency for potentially improved clinical outcomes. Views on patient disclosure and consent were divided: only half believed patients should be informed about MLEA use, while a only a quarter thought patient consent should be required.
These findings diverge from the emerging consensus in ethical discussions that patient consent for MLEA should be required. We conclude that clear ethical guidelines and protocols are needed to navigate transparency tradeoffs and safeguard patient interests in the implementation of ML and other artificial intelligence systems in healthcare, including in the ART clinic.
Keywords: machine learning and assisted reproductive technologies, Patient disclosure and consent, Implementation of novel technologies
Learning Objectives:
After participating in this conference, attendees should be able to:
Understand the role that machine learning and artificial intelligence may play in the assisted reproduction clinic
Assess the importance of patient disclosure and consent in the clinical implementation of machine learning for embryo assessment and how patient interests can be protected
Analyse the implications of the interpretability or opaqueness of machine learning models as they are integrated into clinical practices of embryo assessment and selection, especially for professional and patient trust
.png)
