Using blockchain infrastructure to advance the use of machine learning in the field of ART

Abstract

Research in the assisted reproductive technology (ART) field seeks to address the variety of obstacles to the success of in vitro fertilization (IVF). Artificial intelligence (AI) has the potential to improve efficacy and efficiency of clinical methodologies, hence optimizing treatment. The main focus of existing AI research is on prediction of ART treatment outcome. However, there are still numerous limitations in current AI models such as lack of data standardisation, missing data and bias as most are trained from a single source and evaluated retrospectively. Insufficient datasets result in poor decisions and performance hence AI models cannot currently be applied in clinical practice. Large-scale randomized controlled trials are still required for external validation of these algorithms. A new solution offering big data analytics is needed in order to establish high-quality evidence. The main difficulties with handling these types of data is that they are highly sensitive, strictly regulated, and patient consent and ethical approval is mandatory for research purposes, which makes multi-partner studies on large datasets very difficult. To overcome this challenge, data sharing is vital to allow collaborations compliant with ethical data handling within the confines of international law. Initiating a data hub for fertility using federated learning (FL) would offer decentralised network where algorithms move between clinics, while data never leaves the clinic. The main advantage of this solution is that it solves the problems of privacy and confidentiality, accountability, ethics, security, traceability, non-traveling of data, and learning on multiple datasets, which can be solved by a platform based on blockchain technology allowing federated learning on multiple decentralized datasets. This framework would promote collaborative research and innovation while reducing the cost of innovation validation & implementation.