Comparison Of Machine Learning Algorithms For Cardiotocogram Data

  • Prof. Archana Chaudhari, Madhura Raste, Chinmay Paranjape, Hiten Doshi

Abstract

Cardiotocography is a technique used to monitor fetal health during the gestation period.  The CTG data is conventionally interpreted by experience and knowledge. This makes it prone to bias and human error. Machine learning is increasingly being used in the field of medical sciences. The aim of this study is to identify the most suitable Machine learning algorithm to classify the CTG signals into Normal, Suspect and Pathological hypoxic connditions based on features like accuracy, precision and F1-score and Recall. Further each algorithm was also tested using multiple combinations of characteristics one, with all characteristics; second using the most significant 9 characteristics and third a set of 12 characteristics which included the 9 most significant and 3 less significant characteristics. The performance metrics were derived using the confusion matrix used to interpret the success of the algorithm and set.

Published
2020-07-01
How to Cite
Prof. Archana Chaudhari, Madhura Raste, Chinmay Paranjape, Hiten Doshi. (2020). Comparison Of Machine Learning Algorithms For Cardiotocogram Data. International Journal of Advanced Science and Technology, 29(7), 12437 - 12454. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/27938
Section
Articles