Prediction of gestational success using artificial intelligence through deep learning techniques.

• Grant number: 23/08159-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: September 01, 2023
• End date: December 31, 2024
• Field of knowledge: Health Sciences - Medicine
• Principal Investigator: José Celso Rocha
• Grantee: Bruno Araújo Mendes
• Host Institution: Faculdade de Ciências e Letras (FCL-ASSIS). Universidade Estadual Paulista (UNESP). Campus de Assis. Assis , SP, Brazil

Abstract

Infertility is a worldwide health problem that afflicts a considerable number of people and has several causes, from harmful life habits to genetic factors. This problem attracts many couples to assisted reproduction clinics in order to fulfill their dream of having one or more children. The assisted reproduction technique has been improved over the years in order to increase the pregnancy success rate. Among the advances that have been achieved so far are: the creation of the intracytoplasmic sperm injection technique, the development of incubators such as the EmbryoScope® and Geri®, and the time-lapse technology. Another technology that is gaining space in the research area to assist physicians and embryologists is artificial intelligence. There are several applications of this technique in assisted reproduction, such as the choice and classification of oocytes, sperm and embryos, and even the prediction of implantation and pregnancy rates. Among the best known techniques of artificial intelligence, two stand out: machine learning and deep learning. Machine learning is an approach that involves training algorithms to learn patterns in an attempt to perform specific tasks based on data sets. These algorithms are designed to extract insights from data, identify patterns, and make predictions based on this information. There are different machine learning techniques, such as lineares regression, decision trees, clustering algorithms, and others. Deep learning, on the other hand, is a specific subfield of machine learning that focuses on algorithms known as convolutional neural networks. Convolutional neural networks are inspired by how the human brain works, and are composed of several layers of processing units, called artificial neurons. These networks are capable of automatic learning from large volumes of raw data, identifying complex and hierarchical features in the data. Thus, this research project aims to create and implement a DL- based convolutional neural network that can perform a prognosis regarding the success or failure of pregnancy. All techniques will be developed and implemented in the Matlab® platform, which presents a toolbox specialized in the deep learning process. The project will be developed in the laboratory of applied mathematics of Unesp de Assis - LaMAp, which has sufficient conditions for the development of the research.