Skip to content

Latest commit

 

History

History
66 lines (57 loc) · 3.88 KB

README.md

File metadata and controls

66 lines (57 loc) · 3.88 KB

Fetal-Arc

Predicting Fetal Health, and Birth-Weight of fetus using Machine Learning

Abstract

The moment a child is born, the mother is also born. She never existed before. The woman existed, but the mother, never. A mother is something absolutely new.

The lines talk about pregnancy, which is one of the most beautiful phases in women’s life. To make it nicer and easier, it is utmost important to take care of fetal health.

This project focuses on machine learning techniques used for predicting

  1. Fetal Health as Normal, Suspect or Pathological using cardiotocography (CTG) data
  2. Birth weight of baby using gestational age and mother’s features.

For birth weight prediction, RandomForestRegressor and AdaBoostRegressor are used in a weighted fashion to give final result of birth weight in kgs, with root mean squared error (rmse) being 0.42 on train set and 0.44 on test set. Going for the second problem of predicting fetal health as normal, suspect or pathological, Support Vector Classifier, Decision Tree Classifier, Adaboost Classifier, Random Forest Classifier are used and through majority voting the final label is assigned. This technique gave macro recall score of 0.95 on train set and 0.92 on test set.

Why to address this problem?

Pregnancy is the most delightful period. Healthy pregnancy leads to healthy baby. So fetal care becomes utmost important. According to WHO, one million babies die within 24 hours of birth due to premature birth and complications during birth. Also, around 810 women die each day during delivery or soon after delivery. This really causes the need to take care of fetus with utmost priority.

Problem Background

Cardiotocography (CTG) is well-known and most widely used method to know about fetal health which records (graph) the fetal heartbeat (cardio) and uterine contractions (toco) of mother during pregnancy. It is carried out during third trimester or sometimes even during final trimester and during delivery so as to know if fetal heartbeat is not hampered by uterine contractions. For mother and fetus various parameters are measured which are also captured in dataset. Normal condition occurs when every parameter is within desired range, fetal health is suspectible when one of parameters is abnormal and pathological when more than one parameters are not normal. In case of suspectible, call is made for more tests while for pathological state, there are emergency actions taken by the doctor. The machine learning algorithms provide a quick support and equip the doctors to take actions immediately in case the fetus is in abnormal condition.

Birth weight stands most crucial for fetus, defining the risk during delivery, mortality rate within one year and somewhat related to diseases that occur in adulthood. Birth weight is difficult to measure directly but a rough estimate can be made by experienced doctors. Low birth weight can potentially causes major issues and Over weight can lead to serious injuries to mother and fetus during delivery, hence getting even a rough estimate in this direction would also serve to be of great help. Various machine learning techniques can be employed to predict fetal birth weight by exploiting the features of mother.

Links for dataset

Link for deployed website

https://fetalhealth-simran.herokuapp/

Link for medium article

https://medium.com/analytics-vidhya/fetal-arc-predicting-fetal-health-and-birth-weight-of-the-fetus-using-machine-learning-5d6726323904

For more details, kindly have a look at repository and attached ppt