Uniform distribution tuna swarm optimisation and deep neural network for foetal health classification

Foetal health is generally accessed by foetal heart rate (FHR) monitoring throughout the antepartum period. FHR analysis is a difficult and illogical process because of restricted dependability. Previous research only examined cardiotocographic (CTG) dataset classification accuracy, ignoring computational time, a critical clinical issue. Using uniform distribution tuna swarm optimisation (UDTSO), this paper selects the most important CTG traits. This study developed a machine-learning algorithm to differentiate normal and abnormal foetal CTG data. The proposed study involves pre-processing, FS, classification, and outcomes evaluation. The dataset is normalised using min-max normalisation first in pre-processing. Min-max normalisation modifies characteristics from 0 to 1 range. In the second feature selection step, the UDTSO algorithm selects a subset of input characteristics to evaluate accuracy and choose the optimum solution. Third, a deep neural network (DNN) classifies CTG recordings as normal (N), suspect (S), or pathologic (P). DNN's AlexNet-SVM captures convolution layer filter data. Max pooling minimises weights and concatenates output from a collection of neurons. The fully linked layers now have the AlexNet-SVM classifier to reduce time complexity. Classifiers are assessed on precision, recall, f-measure, and accuracy. The CTG dataset comes from UCI Machine Learning Repository.