/Preterm-Birth-Prediction

This repository contains the code of the paper titled, "Preterm Birth Prediction of Pregnant Women in Post Conization Period Using Machine Learning Techniques", accepted in 11th CSOC 2022, Springer.

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

Preterm Birth Prediction of Pregnant Women in Post Conization Period Using Machine Learning Techniques

Description

Link: https://link.springer.com/chapter/10.1007/978-3-031-09076-9_36

Citation:

Bari, M.A.J., Faisal, M.I., Hasan, M., Islam, L., Hossain, M.S., Momen, S. (2022). Preterm Birth Prediction of Pregnant Women in Post Conization Period Using Machine Learning Techniques. In: Silhavy, R. (eds) Artificial Intelligence Trends in Systems. CSOC 2022. Lecture Notes in Networks and Systems, vol 502. Springer, Cham. https://doi.org/10.1007/978-3-031-09076-9_36

Or, you can download the citation here.

This repository contains code for the implementation of our paper titled Preterm Birth Prediction of Pregnant Women in Post Conization Period Using Machine Learning Techniques, which has been published at the CSOC 2022: 11th Computer Science On-line Conference 2022.

Abstract

Pregnant women who underwent excisional surgeries (conization) for cervical intraepithelial neoplasia (CIN) display high risks of preterm birth. It is crucial to predict the risks of preterm birth amongst women in their post conization periods as this has severe consequences in terms of the cost as well as the health of the mother and the baby. This paper presents a preterm birth prediction system using machine learning approaches which will allow to evaluate the risk of a preterm birth. The dataset used in this work consisted of longitudinal cervical length (CL) of different gestational periods from 725 pregnant women undergoing surveillance programs in three clinics at London University Hospitals. Several machine learning algorithms were applied to make the prediction. Model based on decision tree achieved the highest accuracy (99.3%) on the test dataset.

Dataset used in this work can be downloaded from here.

Proposed Methodology

Acquiring the dataset, we performed adequate data preprocessing so that ML algorithms can efficiently use the data to make predictions.

Proposed Architecture

We have used a handful of machine learning algorithms to empirically find out the best performing algorithm for this dataset. The results of these algorithms have varied in accuracy. The Decision Tree algorithm was found to attain the best accuracy score with 99.3% accuracy on the test data.

Comparison between Machine Learning Algorithms

Authors

  1. Mian Ahmed Jamiul Bari
  2. Mohammad Imtiaz Faisal
  3. Mahmud Hasan
  4. Labiba Islam
  5. Md. Sabbir Hossain