Page 13 - GTM-2-2

Global Translational Medicine Clinical algorithms in ART

decreased ovarian reserve and increased risk of conditions on patient characteristics and treatment data.
such as endometriosis. Another potential explanation v. Random forest: This algorithm is used to analyze
is that women with higher levels of education may be data sets with many variables and identify
more likely to engage in behaviors that can contribute to the most important factors that influence the
infertility, such as smoking, excessive alcohol consumption, outcome. In IVF, random forest can be used to
and exposure to environmental toxins. predict the probability of success or failure based
on patient and treatment characteristics.
On the contrary, education may impact access to
healthcare and reproductive technologies. Women with These algorithms can be trained using large data sets
higher levels of education may have greater knowledge of IVF treatment outcomes, patient characteristics, and
of and access to fertility treatments such as IVF. In treatment data. Once trained, they can be used to predict
comparison, women with lower levels of education may the probability of success or failure for a given patient
face barriers to accessing these treatments, leading to a based on their individual characteristics and treatment
higher risk of infertility. Finally, socioeconomic factors plan. These predictions can help doctors to personalize
such as income and access to healthcare may also play a treatment plans and improve IVF success rates.
role in the relationship between education and female Increasing AI applications in the field of reproductive
infertility. Women with higher levels of education may have medicine to improve the treatment of infertility are already
higher incomes and greater access to healthcare, which can established. Some of the applications are elaborated as follows:
improve their overall reproductive health and decrease i. Oocyte ovarian reservoir estimation: This can
the risk of infertility. While the relationship between help fertility specialists to make more accurate
female educational attainment and infertility is complex, it predictions about the number of oocytes that can
highlights the need for comprehensive reproductive health be retrieved during the IVF cycle.
education and access to fertility treatments for all women, ii. Sperm analysis: AI can analyze and classify sperm
regardless of their educational background. morphology, motility, and concentration with
AI can potentially transform infertility diagnosis and greater accuracy and speed than manual methods,
treatment by enabling more accurate diagnoses, personalized improving the diagnosis of male infertility.
treatment plans, and improved patient outcomes. There are iii. COS: It is done by predicting the optimal dose
several algorithms that can be used to predict the success of and duration of gonadotropin administration.
IVF treatments. Here are some of the most common ones: iv. Fertilization: AI can help embryologists to
i. Logistic regression: This algorithm is used to predict identify the best quality embryos by analyzing
various morphological and kinetic parameters of
the probability of success or failure of IVF treatments the developing embryos.
based on patient and treatment characteristics. It v. Blastulization: AI can help to identify which
uses a mathematical model to analyze data from embryos are more likely to develop into blastocysts,
previous IVF cycles and identify the factors that are improving the success rates of IVF treatments.
most predictive of success or failure. vi. Implantation of human embryo: AI can assist
ii. Decision trees: This algorithm is used to analyze fertility specialists in selecting the best clinical
complex data sets and create decision trees that and embryological parameters to optimize the
map out the most likely outcomes based on implantation rate.
various factors. In IVF, decision trees can be used
to predict the probability of success or failure In the present review, we analyzed the most promising
based on patient age, ovarian reserve, embryo applications to improve results and compliance with IVF
quality, and other factors. procedures, leaving those without clear validation out of
[16]
iii. Neural networks: This algorithm is designed to our focus .
simulate the function of the human brain and IVF programs require a high degree of laboratory
can be used to analyze large and complex data efficiency to optimize outcomes and ensure the safety
sets. In IVF, neural networks can be used to of patients. Several algorithms have been developed and
analyze patient data and identify patterns that are validated to improve the laboratory efficiency of IVF
predictive of success or failure. programs, including:
iv. Support vector machines (SVM): This algorithm i. Time-lapse imaging algorithms: Time-lapse
is used to analyze and classify data based on imaging algorithms use computer vision
complex patterns. In IVF, SVM can be used to techniques to analyze time-lapse images of
predict the probability of success or failure based developing embryos. These algorithms can

Volume 2 Issue 2 (2023) 7 https://doi.org/10.36922/gtm.0308