Page 18 - AIH-2-3

Artificial Intelligence in Health AI in embryo selection for ART

technically demanding than vaginal insemination. Despite focuses on algorithms enabling computers to learn from
its complexity, artificial insemination may be necessary for data, while AI is a vast field that aims to replicate human
specific infertility treatments or breeding initiatives. It allows intelligence through reasoning, problem-solving, and
for the conservation of genetic lines, acceleration of line learning. Although AI is widely used in many sectors, ML
extension, or the synchronization of embryo development by is essential for enhancing decision-making and prediction
enabling one male’s sperm to inseminate multiple females. skills in these applications. AI and ML have the
74
68
This method addresses the issue of male-factor infertility and potential to revolutionize basic science, clinical practice,
can be used with various species, including gorillas, lions, healthcare administration, and medicinal economics. In
70
bears, and tuna fish. Infertility in both males and females reproductive medicine, predictive modeling using AI can
can be successfully treated using artificial insemination, with accurately predict fertility outcomes. However, challenges
a success rate of up to 18.2% every cycle and 58.4% after such as managing large-scale data, identifying valuable
6 months of treatment. 69 features, and validating models with gold-standard study
designs remain. Further precision, standardization,
75
3.8. Relationship between ART and AI and automation in the field of reproductive medicine
A significant portion of people worldwide experience could be achieved using AI-guided procedures. For
76
infertility, which is defined as the inability to achieve example, AI-based ART software can reduce interobserver
a clinical pregnancy following 12 months of regular, variability, personalize drug doses, and improve clinical
unprotected sexual intercourse. 10 Treatment for and operational efficiency, particularly in sperm selection
subfertility, infertility, or genetic disorders that hinder and oocyte quality assessment. Most importantly, AI has
77
natural conception is referred to as ART. Common shown potential in reproductive urology by predicting
ART methods include ET, IVF, ICSI, and GIFT. Recent semen parameters, identifying candidates for genetic
4
technological advancements aim to automate processes testing, and automating sperm detection. In addition, AI
7
such as sperm selection, fertilization, and embryo culture, in ART improves efficiency by identifying patients at risk
which could enhance consistency and reduce the stress for conditions such as endometriosis, detecting gamete
caused by manual manipulation. The integration of AI production values, and optimizing controlled ovarian
13
with ART enhances the effectiveness and success rates stimulation by calculating ideal starting drug doses and
73
of ART treatments. AI algorithms can assess and predict trigger timing using deep learning algorithms. Recently,
the quality of gametes and embryos, which are vital to the Levenberg–Marquardt neural networks trained on local
success of ART. At present, morphological examinations, binary patterns have demonstrated promising outcomes
which are prone to subjectivity and human error, are the in terms of oocyte and embryo quality prediction, offering
primary method used by embryologists to manually assess potential improvements in ART, especially in nations with
the quality of gametes and embryos. By eliminating inter- restricted embryo selection practices. 68
70
observer and inter-objective variations, AI algorithms may
provide a more standardized and objective evaluation. 3.10. Integration of AI in embryo selection
71
This integration enhances reproductive health outcomes In the past, doctors have utilized ML algorithms to assist
by increasing implantation success rates, reducing in selecting embryos for human-assisted reproduction.
the incidence of multiple pregnancies, and improving The challenges of embryo selection have gained significant
single-ET. Moreover, AI can predict embryos’ viability attention with the advent of ART. Invasive techniques,
72
and oocytes’ developmental capability using ML algorithms including preimplantation genetic screening, as well
and morphokinetic parameter analysis. By selecting as transcriptome and proteome analyses of biopsied
the most viable embryos, this method can increase the embryonic tissue, were initially emphasized and are
chances of implantation and successful pregnancy. In currently being explored to obtain direct insights into
73
addition, AI can aid in automating tedious and repetitive embryonic development. In ART, a variety of deep
ART laboratory duties, thereby increasing productivity learning and ML models are applied to enhance embryo
and minimizing errors. However, to achieve extensive selection processes. For example, AI models such as
23
integration into clinical practice, ethical considerations ERICA use blastocyst images to estimate euploidy in
and the imperative for transparency in AI algorithms must embryos. To enhance embryo selection processes in IVF,
78
be duly acknowledged and resolved. AI analyzes complex data, identifies trends, and provides
an objective evaluation of embryos. Furthermore, AI is
79
3.9. Advancements in AI and ML in ART
used to determine the optimal quantity of metaphase II
AI and ML are two related yet distinct disciplines within oocytes required for ART to produce viable blastocysts and
the field of computer intelligence. ML is a subset of AI that embryos. 73

Volume 2 Issue 3 (2025) 12 https://doi.org/10.36922/aih.7170