APPLICATION OF MACHINE LEARNING TO PREDICT RISK OF PRETERM BIRTH IN INDONESIA USING THE 2017 INDONESIA DEMOGRAPHIC AND HEALTH SURVEY DATA
Premature birth is a condition where a baby is born before passing 37 weeks of gestation. Babies born prematurely are at high risk of experiencing health complications and even death due to imperfect organ growth. Premature birth can be caused by various factors, such as medical history and the moth...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/80971 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Premature birth is a condition where a baby is born before passing 37 weeks of gestation. Babies born prematurely are at high risk of experiencing health complications and even death due to imperfect organ growth. Premature birth can be caused by various factors, such as medical history and the mother's socioeconomic conditions. According to USAID 2015, Indonesia is ranked 5th highest in premature births in the world with a prevalence of 15%, where there are 779,000 cases of premature births every year. Therefore, a method is needed that can help predict premature birth, namely by using machine learning-based artificial intelligence.
This research was conducted using the Indonesia National Demographic and Health Survey 2017 dataset. Based on the research conducted, 17 features were obtained which were risk factors, namely those related to the mother's medical history, the mother's smoking activity, and the mother's level of education and knowledge. Then it was found that the Logistic Regression model had the best performance. Model optimization was carried out using logarithmic transformation and hyperparameter tuning so that the premature birth prediction model had an accuracy rate of 73.10%, precision of 75.89%, recall of 67.75%, and ROC-AUC of 80%.
|
|---|