Machine Learning Techniques for Identifying Fetal Risk During Pregnancy

Abstract

Cardiotocography (CTG) is a biophysical method for assessing fetal condition that primarily relies on the recording and automated analysis of fetal heart activity. The quantitative description of the CTG signals is provided by computerized fetal monitoring systems. Even though effective conclusion generation methods for decision process support are still required to find out the fetal risk such as premature embryo, this proposed method and outcome data can confirm the assessment of the fetal state after birth. Low birth weight is quite possibly the main attribute that significantly depicts an unusual fetal result. These expectations are assessed in a constant experimental decision support system, providing valuable information that can be used to obtain additional information about the fetal state using machine learning techniques. The advancements in modern obstetric practice enabled the use of numerous reliable and robust machine learning approaches in classifying fetal heart rate signals. The Naïve Bayes (NB) classifier, support vector machine (SVM), decision trees (DT), and random forest (RF) are used in the proposed method. To assess these outcomes in the proposed method, some of the metrics such as precision, accuracy, F1 score, recall, sensitivity, logarithmic loss and mean absolute error have been taken. The above mentioned metrics will be helpful to predict the fetal risk.

An Elderly Fall Detection System Using Enhanced Random Forest in Machine Learning

“…LSTM and CNN were both used in the experiment by the researchers. The evaluations of performance showed that LSTM excelled in identifying odd behavior in the senior population [10].…”

An Effective Disease Prediction System using CRF based Butterfly Optimization, Fuzzy Decision Tree and DBN

“…A new fuzzy min-max neural classifier with temporal constraints was constructed [17] and also incorporated a met heuristic technique called particle swarm optimization for optimizing the features that are capable of enhancing the classification result on heart, cancer and diabetes datasets. A new neuro-fuzzy classification algorithm was developed [21] for predicting the various diseases such as cancer, diabetes and heart. They have used the bell-shaped fuzzy membership method to perform the fuzzification process on input datasets by applying fuzzification matrix that input patterns were interrelated with classes.…”