Κατεύθυνση Αναπαραγωγική-Αναγεννητική Ιατρική
Library of the School of Health Sciences

2024-04-10

2024

Spiliotis Dimitrios-Emmanouil

Καλαμπόκας Θεόδωρος, Επίκουρος Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Δρακάκης Πέτρος, Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ
Σταύρος Σοφοκλής, Επίκουρος Καθηγητής, Ιατρική Σχολή, ΕΚΠΑ

Εφαρμογές της τεχνητής νοημοσύνης στις τεχνικές της υποβοηθούμενης αναπαραγωγής

Greek

Applications of artificial intelligence in assisted reproductive technology techniques

Artificial Intelligence (AI) is a broad term describing computational systems that
are able to function in a manner similar to that of the human brain and to perform
tasks requiring intelligence. Although the idea of intelligent computing dates back to
the middle of the twentieth century, it was only during the last decade that effort was
being put forth regarding the development of computational models based on artificial
intelligence. The most common methods being utilised for developing such software
are machine learning, deep learning and artificial neural networks.
The digitization of medical records and images has led to a major growth in the
development of artificial intelligence software to be used in various fields of medicine.
Through machine learning and natural language processing, physicians have the
ability to diagnose diseases like chronic kidney disease (CKD) by analyzing the
patient’s record and to perform fully automated analyses of medical images, like MRIs
and PET-CTs, as well as scanned histological slides.
Since the birth of the first child to be born as a result of in-vitro fertilization, the
field of assisted reproduction has made astonishing progress and has seen many
innovative techniques and technological advancements. Nonetheless, the success
rates of ART remain low at an international level. The subjectivity of gamete and
blastocyst selection by embryologists based on morphological parameters is among
the factors accountable for this problem. A solution could be provided through the
integration of AI into the daily practice of IVF laboratories, improving the objectivity
and reproducibility of the assessments carried out, increasing efficiency and
productivity and making complex correlations and predictions impossible for the
human mind to perform.
The main purpose of this thesis is to review existing knowledge about the
applications and benefits of artificial intelligence in assisted reproduction, derived
from international literature. Machine learning, deep learning, artificial neural
networks, support vector and computer vision were found to be the most commonly
used methods. The papers described in the thesis mainly focus on developing and
validating novel computational models, as well as evaluating the clinical importance
of AI software and fully automated analyzers, covering the whole spectrum of ART
techniques, including assessment and selection of gametes and embryos, design of
personalized ovarian stimulation protocols and prediction of IVF success probability
even before the completion of the first cycle.
Studying currently published research, it becomes evident that AI shows immense
potential and holds promises to become an integral part of the new era of assisted
reproduction and contribute to the improvement of IVF success rates. The proposed
algorithms, however, have great room for improvement and optimization and the lack
of randomized clinical trials (RCTs) does not allow for any solid conclusions to be
drawn.

Health Sciences

Artificial Intelligence (AI), Assisted Reproductive Technologies (ART), In-vitro fertilization (IVF), Oocyte assessment, Embryo selection

No

0

Yes

192

107

File access is restricted until 2025-04-11.

https://pergamos.lib.uoa.gr/uoa/dl/object/3396058