Introduction to Machine Learning in Obstetrics and Gynecology

Abstract

This article reviews the history and applications of artificial intelligence and its capacity to support clinical research and practice in the field of obstetrics and gynecology.

Endometrial Cancer Individualized Scoring System (ECISS): A machine learning‐based prediction model of endometrial cancer prognosis

“…The models were created using machine learning algorithms, which carry the privilege of identifying patterns and interactions among potential predictors to create robust models. Machine learning considers all variables and precludes risk of bias secondary to variable selection in traditional statistics and sufficiently treats potential collinearity 8 . The study used a multicenter large database to enhance machine learning algorithms and support the generalizability of results.…”

“…Although these prognostic factors are not included in FIGO staging system, they are growing evidence that, at least some of Machine learning considers all variables and precludes risk of bias secondary to variable selection in traditional statistics and sufficiently treats potential collinearity. 8 The study used a multicenter large database to enhance machine learning algorithms and support the generalizability of results. Unlike FIGO staging, which is a surgical staging that should be fully determined after the procedure, the Endometrial Cancer Individualized Scoring System (ECISS) model I employs preoperative data, and model II provides the option of testing a treatment plan based on preoperative assessment.…”

“…Machine learning is a subtype of artificial intelligence where the machine progressively recognizes patterns that link variables to a specific outcome. In comparison to traditional statistics, machine learning establishes comprehensive individualized prediction models rather than recognizing associations between individual variables and a clinical outcome 8 . The objective of this study is to develop an EC scoring system that predicts individualized prognosis based on patient demographics and disease characteristics using machine learning algorithms.…”

“…In comparison to traditional statistics, machine learning establishes comprehensive individualized prediction models rather than recognizing associations between individual variables and a clinical outcome. 8 The objective of this study is to develop an EC scoring system that predicts individualized prognosis based on patient demographics and disease characteristics using machine learning algorithms. In addition, the study aims to predict prognosis in response to a therapeutic plan, which would facilitate individualized treatment decisions in women with EC.…”

Use of artificial intelligence and deep learning in fetal ultrasound imaging

“…DL models are able to analyze large amounts of data, in a layered, non‐linear manner, using pattern recognition to extract highly representative image features in order to ‘label’ an image (e.g. as normal or abnormal) 14 . DL models can be developed using a supervised or unsupervised learning approach 6 .…”