Application

Summary of Application

Relevant Studies using AI for Application

Fetal heart monitoring (FHR)

FHR is vital information when assessing the health status of a fetus. Its use is a part of the standard procedure for intrapartum assessment for fetal well-being and aids in diagnosing cardiac disorders [23] .

Computer model CAFE (computer-aided fetal evaluator)— a hybrid system that recorded the FHR and uterine contraction signals and proceeded to make decisions as an expert using artificial neural networks.

The system could also detect sudden changes in the baseline, highlighting anomalies to physicians [24] .

Computer-Aided Diagnosis (CAD) systems—uses CNNs that process signals from a database and self-learn essential features from the FHR signal for fetal asphyxia.

This system’s self-learning ability allowed for a higher accuracy rate as the network was able to automatically extract features from the signal, which allowed for less loss of information [25] .

Gestational diabetes mellitus (GDM)

GDM occurs when carbohydrate intolerance is recognized for the first time during pregnancy.

Obesity, maternal age, and miscarriages are among the risk factors associated with women who have GDM.

For pregnant mothers, the United States Preventive Services Taskforce recommends screening protocols for GDM, which can often be overlooked for pregnant women that live in low to middle-income areas or countries due to the lack of availability [26] .

Online Calculator Screening tool for GDM—uses patient risk factors such as high blood pressure, smoking, weight, diet, and ethnicity. Calculates risk for GDM using ANN and backpropagation methods. Patients upload data from surveys, and the calculator outputs the GDM risk factor. Polak and Mendyk (2004) used this method on the record of 2551 women, of which 91 were diagnosed with GDM. The calculator had an accuracy of 70% where it detected true positive GDM—but with more data and development of the ANN, accuracy would increase and become more economically efficient [27] [28] .

In vitro fertilization (IVF) treatment

Accurate and early prediction of the outcome of in vitro fertilization (IVF) treatment is an essential tool for patients and physicians as patients want to know their outcomes before deciding to undergo the emotional and physical toll of conceiving, and it is difficult for physicians to predict the possible outcome intuitively [29] .

Prediction using data mining and decision tree algorithms— data mining allows for the extraction of previously unknown meaningful information from data. This technique is used along with decision trees to determine significant attributes that lead to IVF treatment predictions.

A study used 28 attributes to build a prediction model that was not only able to predict the outcome but also help physicians tailor the treatment for the patient precisely [30] .