In Vitro Fertilization:
A study done by Guh et al. used data mining and AI to create a computer
algorithm that can help predict pregnancy in vitro fertilization (IVF)
[20]. Data mining (DM) uses AI in conjunction with advanced
statistics to discover patterns in large databases. DM extracts the
information of interest and is also able to find new key elements that
might influence the outcome, thus increasing the amount of data that can
be utilized [13,21]. Finding new trends that influence success rates
for IVF are important for the patient and the clinician, to have
realistic outcome expectations. To help clinicians predict pregnancy
success rates, they created a hybrid intelligence model that used DM to
integrate genetic algorithm-based and decision tree learning techniques
that extracted information from the IVF patient records. They found that
this model not only helped predict outcomes but suggested modified IVF
treatment according to individual patient characteristics. A downside of
the study is that the model they created used data from only one IVF
center. However, if the centers united to share information, their
pooled data can significantly expand to represent a wider population
with increased accuracy. Other studies suggest using ANN systems to
predict IVF outcomes by using Learning Vector Quantizer which allows
generalization and standard parameters for enhanced predictive power
[20]. Another aspect of IVF and AI is the possibility of identifying
the most viable oocytes and embryos. An AI system used by Manna et al.
suggests combining AI to extract texture descriptors from an image
(local binary pattern), and assembling it using an ANN [22]. These
results proved to be above average when compared to current methods and
can help to select the best possible oocytes or embryos noninvasively
and objectively. Furthermore, they highlight the advantages of this
technology in selecting the most viable embryos, even in countries where
legislation precludes embryo selection by sex.