Machine learning based Maternal Risks and Fetus health Monitoring System

CHAPTER 1
INTRODUCTION
1.1 Introduction:

Pregnancy complications may be caused by conditions women have before pregnancy or

conditions women develop during pregnancy. The impact of pregnancy complications on maternal
and neonatal outcomes is difficult to estimate because pregnancy complications are made up of a
broad range of conditions with varying levels of severity. Every year, an estimate of 358,000
maternal mortalities is recorded worldwide, with about 99% cases occurring in poor developing
countries. Early screening is expected to reduce maternal mortality rates. The purpose of this
research, therefore, was to analyze the risk factors associated with risk level in the dataset, and to
also identify fit/final models with the capacity to predict maternal complications.
We will train a machine to learn a process from a dataset without programming, as the term
"Machine learning" suggests. Machine learning has started to have a personality of its own
because of its capacity for self-learning as well as other considerations. Reinforce learning,
unsupervised learning, and supervised learning are the three main subcategories of machine
learning. In the areas of handwriting, face, human mood swings, voice, pattern recognition, etc.,
deep learning is now outperforming all other technologies.
Fetals are delivered f rom the women’s womb and before that during the pregnancy the
information about the fetus is tough to get. We can only get the information that there is a fetus
and it would be delivered. So, as the information is not readily available the obstetricians who
check the condition of the fetal rely on the indirect information. One of the most dependent
information is about the fetal heart rate and there is a restriction of electronic fetal heart
monitoring that this variation is hindering the records of the accurate communication and time
management. There is an another component known as the cardiotocogram which contains distinct
signals and is mainly used for recordings of the fetal heart rate which is the main way through
which the obstetricians rely on the information. But the trend seen in these days by the doctors is
that there is very high intra and inter observer fluctuation in FHR patterns. But there is a risk in
which a falsely diagnosed fetal pain may lead to unnecessary interventions. Hence, the main
motive of the research is to employ machine learning algorithms to classify the methods as there
can be a room for error by the doctor but the prediction algorithm may perform really well in this
case and also help in monitoring the results and give a proper analysis than the doctor could get by
his own or someone’s observation.

Dept.of CS&E,BIET,Davangere 1
Machine learning based Maternal Risks and Fetus health Monitoring System

Pregnancy involves various risks for both the woman and the fetus. Machine learning (ML)
techniques offer valuable tools for detecting and classifying these risks, enhancing early
intervention and improving outcomes.
Maternal Health Risks:
● Possible Risks: Gestational diabetes, preeclampsia, maternal hypertension.
● ML Application: Machine learning models can analyze maternal health data,
identifying patterns and predicting the likelihood of these conditions, allowing for timely
monitoring and intervention.
Fetal Health Risks:
● Possible Risks: Intrauterine growth restriction, congenital anomalies, fetal distress.
● ML Application: ML algorithms applied to fetal monitoring data, such as
cardiotocography (CTG), can assist in early detection of abnormalities, enabling healthcare
professionals to intervene and manage risks proactively .
Pregnancy Complications:
● Possible Risks: Preterm birth, stillbirth, multiple pregnancies.
● ML Application: ML models can analyze historical data to predict the likelihood of
complications, allowing healthcare providers to implement preventive measures and
personalized care plans .
● ML Application: ML algorithms can assess various factors to predict the mode of
childbirth, aiding in decision-making and reducing the risk of complications associated with
specific delivery methods.
In summary, machine learning provides a data-driven approach to analyze complex relationships
within maternal and fetal health datasets, enabling accurate risk prediction and classification.
This empowers healthcare professionals with timely information for informed decision-making,
ultimately improving outcomes for both the pregnant woman and the fetus.

Dept.of CS&E,BIET,Davangere 2
Machine learning based Maternal Risks and Fetus health Monitoring System

CHAPTER 2
LITERATURE SURVEY
2.1 Literature Survey Review:

The term Severe Maternal Morbidity (SMM) includes a set of complications that can have
a severe adverse effect on women and baby health, and happen during pregnancy, delivery. When
any of these appear, it is necessary to provide the patient with immediate attention, in order to
avoid death. Although maternal health outcomes have shown positive variation, complications of
pregnancy still are an important public health issue. Each year around 585.000 women die during
pregnancy, delivery and annually close to 50 million complications in maternal health are
registered, and approximately 300 million women suffer from short and long-term illnesses and
injuries related to pregnancy, childbirth and postpartum.

Md Assaduzzaman; Abdullah Al Mamun; Md Zahid Hasan “Early Prediction of Maternal

Health Risk Factors Using Machine Learning Techniques” [1] In rural areas, pregnant women
face various difficulties and challenges, including a shortage of doctors, inadequate knowledge, a
lack of public clinics, infrastructure issues, and transportation issues. The mother’s pregnancy is
the major cause of the infant’s poor health, rather than any other factors that may have arisen
after childbirth. Significant roles are played by maternal risk factors such as the mother’s chronic
condition, age, nutrition, and other medical assistance during pregnancy. Recent developments in
Artificial intelligence methods, particularly machine learning models, have made it easier to
make predictions in a variety of disciplines. We can identify the primary maternal risk factors
that can lead to newborn child and maternal mortality using machine learning techniques. This
paper proposes improved data preprocessing methods that involve feature engineering and data
cleaning in order to effectively handle anomalies in the data values. To identify the maternal
health risk factor, several machine learning algorithms were used, including Cat Boost, Random
Forest, XGB, Decision Tree, and Gradient Boost. Using the preprocessed dataset, the suggested
model was developed, trained, and tested. The Random Forest was the best machine-learning
algorithm with an accuracy score of 90%.

Islam MN, Mustafina SN, Mahmud T, Khan NI: “Machine learning to predict pregnancy
outcomesYear:2022” [2] A systematic review, synthesizing framework and future research
agenda. BMC Pregnancy Childbirth .Machine Learning (ML) has been widely used in predicting
the mode of childbirth and assessing the potential maternal risks during pregnancy. The primary
Dept.of CS&E,BIET,Davangere 3
Machine learning based Maternal Risks and Fetus health Monitoring System

aim of this review study is to explore current research and development perspectives that utilizes
the ML techniques to predict the optimal mode of childbirth and to detect various complications
during childbirth. A total of 26 articles (published between 2000 and 2020) from an initial set of
241 articles were selected and reviewed following a Systematic Literature Review (SLR)
approach. As outcomes, this review study highlighted the objectives or focuses of the recent
studies conducted on pregnancy outcomes using ML; explored the adopted ML algorithms along
with their performances; and provided a synthesized view of features used, types of features, data
sources and its characteristics. Besides, the review investigated and depicted how the objectives of
the prior studies have changed with time being; and the association among the objectives of the
studies, uses of algorithms, and the features.

Naveen Reddy Navuluri: “Fetal Health Prediction using Classification Techniques”

Year:2021[3]A fetal is basically an unborn offspring which is in the embryo stage until it comes to
the world. During the pregnancy process, each three month period is known by a name called
trimester. During this process the fetus grows and develops and along with it the regular checkups
are very important. As we all know that a pregnancy lasts for 9 months and in this long period
there may be various reasons which may cause disability or mortality in the newborn which is a
very severe case and this needs to be avoided. One of the main tool to analyze the health of the
fetal in the womb is by doing a CTG(Cardiotacagraphy) which generally is used to evaluate the
heart beat and the uterine contractions hence the data generated is used by the doctor to analyze
the health and give his wording. But there is a room for error hence the doctors are not reliable to
analyze the data hence different machine and deep learning algorithms have been there which can
analyze the data and predict the fetal health based on it. The main motive of the paper is to prove
the prediction accuracy using the different classification models and compare which model
performs better.

Zheng T, Ye W, Wang X, Li X, Zhang J, Little J: “A simple model to predict risk of gestational

diabetes mellitus from 8 to 20 weeks of gestation in Chinese women. BMC Pregnancy
Childbirth.”Year:2019 Gestational diabetes mellitus (GDM) is associated with adverse perinatal
outcomes. Screening for GDM and applying adequate interventions may reduce the risk of adverse
outcomes. However, the diagnosis of GDM depends largely on tests performed in late second
trimester. The aim of the present study was to bulid a simple model to predict GDM in early
pregnancy in Chinese women using biochemical markers and machine learning algorithm.Data on
a total of 4771 pregnant women in early gestation were used to fit the GDM risk-prediction model.

Dept.of CS&E,BIET,Davangere 4
Machine learning based Maternal Risks and Fetus health Monitoring System

Predictive maternal factors were selected through Bayesian adaptive sampling. Selected maternal
factors were incorporated into a multivariate Bayesian logistic regression using Markov Chain
Monte Carlo simulation. The area under receiver operating characteristic curve (AUC) was used
to assess discrimination.

Dept.of CS&E,BIET,Davangere 5
Machine learning based Maternal Risks and Fetus health Monitoring System

2.1 Literature Review Summary:

Study Title Key Classification Challenges Faced

Techniques
Used
Islam MN, Predict ML Support Imbalanced
Mustafina SN, pregnancy techniques Vector datasets, where
Mahmud T, outcomes: a for Machines and one class is
Khan NI (Year: systematic predicting Neural more prevalent
2022) review, childbirth Networks. than another can
synthesizing mode and lead to biased
framework and assessing models.
future research maternal
agenda. risks during
pregnancy
Naveen Analyzing Fetal CTG Different Room for error
Reddy Health Prediction (Cardiotocagra classification in doctor
Navuluri Accuracy Using phy), different models analysis of
(Year: 2021) Machine Learning machine and CTG data
and Deep deep learning
Learning algorithms
Algorithms
Zheng T, Ye A simple model Biochemical Bayesian Diagnosis of
W, Wang X, Li to predict risk markers, logistic GDM depends
X, Zhang J, of gestational machine regression on tests in late
Little J (Year: diabetes learning second
2019) mellitus from 8 algorithm trimester
to 20 weeks of
gestation in
Chinese
women.
Deep hybrid Taofeeq Hybrid model Random forest Limited dataset
model for Oluwatosin Algorithim details and
maternal health Togunwa absence of key
risk variables may
classification in impact the
pregnancy: model's
synergy of ANN generalizability
and random
forest
A. K. Pradhan, A machine SMOTE, Random Forest Algorithm-
J. K. Rout, A. learning RFE dependent
B. Maharana, approach for (Recursive accuracy
B. K. the prediction Feature variations
Balabantaray, of fetal health Elimination)
and N. K. Ray using CTG

Dept.of CS&E,BIET,Davangere 6
Machine learning based Maternal Risks and Fetus health Monitoring System

2.2 Existing System:

The current system, relying on traditional methods such as Cardiotocography (CTG) for assessing
fetal health during pregnancy, is susceptible to potential errors and high variability in interpreting
Fetal Heart Rate (FHR) patterns. This variability may result in false alarms regarding fetal
distress, potentially prompting unnecessary interventions. Diabetes screening usually involves a
Glucose Challenge Test (GCT) Test at 24-28 weeks. If elevated or with risk factors, then further
tests are conducted.

2.3 Problem Statement:

Maternal health during pregnancy is a critical concern, and predicting associated risks is vital for
ensuring the well-being of both the mother and the fetus. Traditional methods may lack precision
and early detection capabilities. The problem lies in the need for a more accurate, proactive, and
data-driven approach to assess maternal health risks and monitor fetal well-being throughout
pregnancy. Current systems may not effectively leverage advancements in machine learning and
analytics to analyze diverse health data and identify subtle risk factors. The challenge is to develop
a robust machine learning-based system capable of predicting maternal health risks and fetal
health status accurately.

2.4 Proposed System:

The system employs advanced machine learning algorithms, including Naïve Bayes, K-Nearest
Neighbor, Decision Tree, and Random Forest, to develop a robust predictive model for maternal
and fetal health assessment. The goal is early identification, enabling timely interventions for
improved outcomes in maternal and fetus health. Users can input maternal and fetal health
attributes through the graphical user interface, and the deployed model provides predictions
regarding potential complications during pregnancy.

Dept.of CS&E,BIET,Davangere 7
Machine learning based Maternal Risks and Fetus health Monitoring System

2.5 Objectives:

 To identify and extract essential features from the maternal and fetal health datasets by
using advanced preprocessing methods, addressing challenges such as missing data and
outliers.
 To design and develop a robust machine learning model for accurately classifying maternal
health risks, utilizing ML algorithms
 To design and implement an intuitive Graphical User Interface (GUI) that enables
healthcare professionals to interact seamlessly with the machine learning model, facilitating
easy interpretation of risk assessments.

Dept.of CS&E,BIET,Davangere 8
Machine learning based Maternal Risks and Fetus health Monitoring System

CHAPTER 3
SYSTEM REQUIREMENTS AND SPECIFICATION
3.1 SOFTWARE REQUIREMENTS
This section specifies the minimum software configuration required for the system. Operating
System :
 Windows 7 or any compatible
 Programming language: Python
 IDE : Visual studio.
 Libraries : Pandas, NumPy, sklearn, Seaborn etc.

3.2 HARDWARE REQUIREMENTS

This section specifies the minimum hardware configuration which is needed for the system.
 Processor: Intel Core i3 or advanced
 RAM: 8 GB or Higher
 Hard disk: 500 GB or more

Dept.of CS&E,BIET,Davangere 9
Machine learning based Maternal Risks and Fetus health Monitoring System

CHAPTER 4
METHODOLOGY

Hypertension Diabetes

Feature selection

Data Pre-processing

KNN

Decsion Tree

Naive Bayes

Random
Forest

Fig 4.1: Methodology of Maternal Risks

Dept.of CS&E,BIET,Davangere 10
Machine learning based Maternal Risks and Fetus health Monitoring System

Split Dataset
Feature Extraction Into train set &
test set

Trained Models

Fig 4.1.1 Methodology of Fetal Health Classification

In this project, the methodology is structured into three essential stages. The initial phase involves
data preparation, where the Fetal Health and maternal dataset is utilized. Null values are
systematically removed to enhance dataset quality, ensuring the accuracy of subsequent analyses
and model training. Following data preparation, the feature engineering phase begins with a
comprehensive dataset description, providing foundational insights. These feature engineering
activities aim to provide a deeper understanding of the dataset and its characteristics to contribute to
selecting relevant features for model training.

Dept.of CS&E,BIET,Davangere 11
Machine learning based Maternal Risks and Fetus health Monitoring System

In our proposed method, maternal risk and fetal classification can be detected more efficiently and
less costly within a short time. The working of the system starts with collection of data and
selecting important attributes. Then the required data is preprocessed into the standard format. In
the first stage, preprocessed data are divided into two parts. Most of those are used in the training
phase (80%), and the rest (20%) are used in the testing phase. The proposed system has trained our
dataset using machine learning algorithms like Decision tree, Random Forest, Naive Bayes and
KNN. Using Visual studio platform, we have trained and finally predicted the result of a patient.
The accuracy of the system is obtained by testing the system using the testing data.
This system is implemented using the following modules.
 Dataset Collection
 Data Preprocessing
 Feature selection
 Splitting Dataset
 Classification models

4.1 Dataset Collection

Initially, we collect datasets from the Kaggle website for maternal risk prediction and fetal
classification system. The dataset consists of 6 features and a total of 1014 patient records which
are used for predicting maternal risk and the dataset consists of 22 features and a total of 2126
patient records which are used for fetal health classification. We utilize a dataset comprising
individuals who have conducted analyses and tests for identifying maternal risks and classifying
fetuses. In the dataset, patients are represented by rows, while factors or attributes (features)
under examination are represented by columns, forming a matrix structure.
 Age: Age in years when a woman is pregnant.
 SystolicBP: Upper value of Blood Pressure in mmHg, another significant attribute during
 pregnancy.
 DiastolicBP: Lower value of Blood Pressure in mmHg, another significant attribute during
 pregnancy.
 BS: Blood glucose levels is in terms of a molar concentration.
 BodyTemp: Pregnant women’s body temperature.
 HeartRate: A normal resting heart rate in beats per minute.
Dept.of CS&E,BIET,Davangere 12
Machine learning based Maternal Risks and Fetus health Monitoring System

 Risk Level: Predicted Risk Intensity Level during pregnancy considering the previous
attribute.
4.2 Pre-processing of data
Data pre-processing plays a vital role in the development of a machine learning model.
Initially, the data may be in an untidy or incompatible format, leading to potentially erroneous
outcomes. During the data pre-processing phase, we convert the data into the necessary format.
Ensuring the dataset's reliability and accuracy, this step becomes crucial in addressing factors
like noise, duplicates, and missing values present in the data. Data pre-processing encompasses
a range of activities that include collecting datasets, partitioning datasets, data cleaning,
factorization, missing values imputation, etc. The purpose of data pre-processing is to enhance
the accuracy of the model by refining the data before it is used for analysis or modeling.

Fig 4.2 : Maternal Dataset

• Data Cleaning:
NA values in the dataset is the major setback for us as it will reduce the accuracy of the
prediction profoundly so, we will remove the fields which does not have values. We will
substitute it with the mean value of the column. This way, we will remove all the values in the
data set.
• Data Integration:
Data Integration is a data preprocessing technique that combines data from multiple
heterogeneous data sources into a coherent data store and provides a unified view of the data.
While implementing data integration, it should work on data redundancy, inconsistency,
duplicity, etc.
• Data transformation:
Data transformation is an essential data preprocessing technique that must be performed on the
Dept.of CS&E,BIET,Davangere 13
Machine learning based Maternal Risks and Fetus health Monitoring System

data before data mining to provide patterns that are easier to understand. Data transformation
changes the format, structure, or values of the data and converts them into clean, usable data.
• Data Reduction:
Data reduction is a process that reduces the volume of original data and represents it in a much
smaller volume. Data reduction techniques are used to obtain a reduced representation of the
dataset that is much smaller in volume by maintaining the integrity of the original data.

4.3 Feature Selection

In a data set where we have a large set of features. Feature selection includes the selection of
appropriate attributes for the prediction system. To detect the links between attributes, we only
choose those attributes that are highly dependent each other in order to apply Machine Learning
algorithms and achieve better accuracy.
Below are some benefits of using feature selection in machine learning:
 It helps in avoiding the curse of dimensionality.
 It helps in the simplification of the model so that it can be easily interpreted by the
researchers.
 It reduces the training time.
 It reduces overfitting hence enhance the generalization.
There are mainly two types of Feature Selection techniques, which are:
 Supervised Feature Selection technique: Supervised Feature selection techniques
consider the target variable and can be used for the labelled dataset.
 Unsupervised Feature Selection technique: Unsupervised Feature selection techniques
ignore the target variable and can be used for the unlabelled dataset.

4.4 Splitting Dataset

After the collection of the dataset, we split the dataset into training data and testing data. The
accurate classification result of the dataset depends on the training and testing phase.

4.5 Classification Models

4.5.1 Decision Tree Decision
Tree is a versatile supervised learning method applicable to classification and regresson tasks,
with a primary focus on solving classification problems. It takes the form of a tree-like
structure, where internal nodes correspond to dataset features, branches represent decision
rules, and each leaf node signifies an outcome. Within the Decision Tree, two key types of
Dept.of CS&E,BIET,Davangere 14
Machine learning based Maternal Risks and Fetus health Monitoring System

nodes exist: the Decision Node and the Leaf Node. Decision nodes in a decision tree are
responsible for making decisions and have multiple branches leading to different paths. In
contrast, leaf nodes represent the final output of those decisions and do not contain any further
branches. To predict the class of a given dataset using a decision tree, the algorithm initiates
from the root node of the tree. It compares the values of the root attribute with the
corresponding attribute in the dataset and, based on this comparison, follows the appropriate
branch to the next node. At the next node, the algorithm once again compares the attribute
value with the sub-nodes' values, thus progressing further along the tree. This iterative process
continues until the algorithm arrives at the leaf node, which provides the final prediction for the
dataset.

4.5.2 K-Nearest Neighbor (KNN)

K-Nearest Neighbor (K-NN) represents one of the simplest supervised machine learning
algorithms. It operates by storing all available data and classifying a new data point based on its
similarity to existing data points. This means that when new data emerges, K-NN can readily
categorize it into a suitable class based on the algorithm's stored data. While K-NN can be
employed for both regression and classification tasks, it is primarily used for solving
classification problems. An essential characteristic of K-NN is its non-parametric nature, as it
does not assume any specific underlying data distribution. Additionally, it is often referred to as
a "lazy learner" algorithm because it postpones learning from the training set until the time of
classification. During the training phase, K-NN simply retains the dataset. When presented with
new data, it categorizes it into a class that closely resembles the characteristics of the new data
point.
4.5.3 Naive Bayes Naive Bayes
This algorithm is a supervised learning technique grounded in Bayes theorem, specifically
designed to tackle classification problems. It stands as one of the straightforward yet highly
effective Classification algorithms, renowned for constructing rapid machine learning models

capable of swift predictions. Its primary application lies in text classification, especially in
scenarios with high-dimensional training datasets. Naive Bayes operates as a probabilistic
classifier, making predictions based on the likelihood or probability of an object's occurrence.
4.5.4 Random Forest:
Random Forest is a highly popular machine learning algorithm utilized within the realm of
supervised learning. It proves effective for addressing both Classification and Regression
challenges in the field of Machine Learning. The foundation of this algorithm lies in the
Dept.of CS&E,BIET,Davangere 15
Machine learning based Maternal Risks and Fetus health Monitoring System

concept of ensemble learning, where multiple classifiers are combined to tackle intricate

problems and enhance model performance. As a classifier, Random Forest comprises numerous
decision trees created from diverse subsets of the provided dataset. To enhance predictive
accuracy, it takes the average of the outputs from these decision trees. The inclusion of a
greater number of trees within the forest enhances accuracy while also mitigating the risk of
overfitting. The Random Forest operates in two phases: First, it constructs the random forest by
amalgamating N decision trees, and second, it makes predictions for each tree created during
the initial phase.

Dept.of CS&E,BIET,Davangere 16