Content uploaded by Rabia Riaz
Author content
All content in this area was uploaded by Rabia Riaz on Sep 17, 2020
Content may be subject to copyright.
Available via license: CC BY 4.0
Content may be subject to copyright.
(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 12, 2019
120 | P a g e
www.ijacsa.thesai.org
A Comparative Study of Supervised Machine
Learning Techniques for Diagnosing Mode of
Delivery in Medical Sciences
Syeda Sajida Hussain1, Rabia Riaz3
The University of Azad Jammu and Kashmir
Muzaffarabad, 13100
Pakistan
Tooba Fatima2
SQA Engineer, DataCheck Limited
Mezzanine Floor Bahria Complex III
M T Khan Road Near American Embassy
Karachi, Pakistan
Sanam Shahla Rizvi4
Raptor Interactive (Pty) Ltd, Eco Boulevard
Witch Hazel Ave, Centurion 0157, South Africa
Farina Riaz5
Independent Researcher, Australia
Se Jin Kwon*6
Dept. of Computer Engineering
Kangwon National University, 346 Joongang-ro
Samcheok-si, Gangwon-do 25913, Korea
Abstract—The uses of machine learning techniques in medical
diagnosis are very helpful tools now-a-days. By using machine
learning algorithms and techniques, many complex medical
problems can be solved easily and quickly. Without these
techniques, it was a difficult task to find the causes of a problem
or to suggest most appropriate solution for the problem with high
accuracy. The machine learning techniques are used in almost
every field of medical sciences such as heart diseases, diabetes,
cancer prediction, blood transfusion, gender prediction and
many more. Both supervised and unsupervised machine learning
techniques are applied in the field of medical and health sciences
to find the best solution for any medical illness. In this paper, the
implementation of supervised machine learning techniques is
performed for classifying the data of the pregnant women on the
basis of mode of delivery either it will be a C-Section or a normal
delivery. This analysis allows classifying the subjects into
caesarean and normal delivery cases, hence providing the insight
to physician to take precautionary measures to ensure the health
of an expecting mother and an expected child.
Keywords—Machine learning; supervised learning;
bioinformatics; medical sciences
I. INTRODUCTION
Bioinformatics is now-a-days the most important field that
is associated with the concepts of machine learning. Almost
every main medical problem can now be solved by
implementing the machine learning techniques such as
classification, regression analysis, clustering, etc. [1] [2] [3].
Both supervised and unsupervised machine learning
techniques can be implemented on medical datasets, based on
the nature of the data and the type of results to be inferred
from the data.
There is a huge corpus of data available for applying
machine learning techniques related to the medical problems.
In almost every field of Bioinformatics, the techniques of
machine learning are implemented and providing very helpful
results for the diagnosis of the disease. The use of these
techniques is very helpful for the medical technicians and
doctors as well as practitioners to correctly perform the
treatment of a disease. These techniques are also supportive
for the future researchers to devise more ways of solving the
problems related to any medical issue effectively. In short,
machine learning has provided a new life span to the field of
Bioinformatics for solving medical related issues.
Many datasets are also available online about the domain
of maternity related cases. Many different types of judgements
can be performed related to those datasets, such as gender
prediction of a child, weight of new born baby, mode of
delivery of the baby and many more. The correct prediction
about the birth mode of a child is important, not only for the
survival of the new born but also for the health of a becoming
mother. So, decision about the mode of delivery of a woman
should be carried out very carefully. In this research, a
medical dataset consisting of the real-world values from the
medical records of the pregnant women has been obtained for
evaluating machine learning techniques to select the most
appropriate technique for solving such type of problems.
Machine learning is a scientific discipline which focuses
on how machines learn from the given data. Machine learning
is a field of artificial intelligence, that provide a system of
automated learning and producing the desired outcome from
the given dataset based on the previous examples from the
same domain.
Samuel, the father of machine learning term divided it in
Supervised and Unsupervised categories [4]. In supervised
learning, we have a training data, with a defined set of rules.
Based on those rules, the testing data will be evaluated. The
*Corresponding Author
(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 12, 2019
121 | P a g e
www.ijacsa.thesai.org
main goal of supervised machine learning is to predict a
known output or target from a huge volume of the input data.
Because of these predictions, the evaluation of the learning
methods will be performed by classifying some metrics.
Supervised machine learning techniques are very important
for performing Classification, Inference or Regression
analysis on a set of data [5] [6] [7]. A study [8] discussed that
a supervised machine learning model is built by dividing a
dataset into two parts: One set is used for building a
classification model by assigning every attribute to one of the
defined class labels. The other is for testing the classification
model.
In unsupervised learning, there are no group boundaries
defined and patterns are matched and recognized from the data
that has no labels for identification of the data. Unsupervised
learning is a type of machine learning algorithm used to draw
implications from datasets consisting of input data without
labelled responses. The most common unsupervised learning
method is cluster analysis, which is used for exploratory data
analysis to find hidden patterns or grouping in data.
The main motivation for this research is to provide aid to
the pregnant women for getting their new born baby healthier
and in a safe way. No doubt, the maternity related issues are
very delicate to handle and solve as, it is the matter of two
lives and a concern of a whole family. So, in this research, the
focus is to perform some affective work for the betterment of
expecting women and to help the medical officers for
performing the maternity related delicate decision about the
mode of delivery very carefully and correctly.
The problem to be solved in this research work is about the
classification of the data of pregnant women in to Caesarean
Section or Normal, identifying their mode of delivery based
on the number of attributes describing different aspects of a
pregnant woman. The Supervised machine learning techniques
of Classification are applied on the sampled dataset for
assigning a class label of either 0 or 1 to an expecting woman.
This classification will be done by using the techniques of
Neural Networks, Support Vector Machines (Linear) and Tree
based classification (Random Forest, CTree, RPART).
The rest of the paper is organized as follows. Related
previous work is reviewed in Section II. Section III describes
the data used in the research and the techniques and tools
adopted to perform the classification techniques. The results
of the experiments in the form of accuracy metrics, ROC
curves and graphs are discussed in Section IV. Section V
provides conclusion and recommendations for the future work.
II. LITERATURE REVIEW
There is a huge volume of research work performed to
determine relationship between C-Sections and inter related
risk factors. Defined maternal age of above 44 has more
chances of medical complications [9] e.g. hypertension and
diabetes and results in higher rate of C-Section delivery. They
used machine learning techniques [10] to find if high blood
pressure and pulse rate, lack of education and low income,
previous surgery and multivitamins are causes of C-Section
delivery. Hueston defined site to site variations have chances
of C-Section delivery [11]. They said that patients of age
greater than or equal to 35 cannot effect in C-Section delivery
but if weight is greater than 3600gm and age is also greater
than 35 than there are more chances of C-Section delivery [12].
A research about C-Section delivery in 2015 reported that
wealth and education effect C-Section delivery [13]. 23 to 35
per cent C-Section deliveries occur in high income people and
12 per cent C-Section deliveries of low income group. Same
as higher education have higher C-Section deliveries, the
women with no education have C-Section birth rate 7.5 and
the metric, secondary and higher education have 21, 31 and 41
per cent, respectively. If there was a previous C-Section birth,
then there are more chances of C-Section birth.
Another research described that many factors are involved
in C-Section [14]. Major factors for C-Section delivery
showed the relationship between section birth, wealth,
education, age, ultrasonography, pregnancy completions.
Private hospitals have more C-Section delivery as compare to
the public. A study said that maternal age affects the C-
Section delivery [15]. The authors divided the age in 3 groups.
One group is less than 35, another is between 35 and 39 and
one group is above 40. Authors defined that the age of 35 -39
have increasing risk of miscarriage and felt chromosomes
abnormalities and age of 40 and above have risk factors of
gestational diabetes, placenta Persia, placenta abrupt and C-
Section delivery.
A reasonable amount of work has been performed on the
coeliac disease effect on the new born [16]. They studied that
if father suffered from any coeliac disease, new born have
lower birth weight and shorter pregnancy duration. If mother
suffer from coeliac then birth weight of new born will also be
low. The factors of cigarette smoking and hypertension during
pregnancy increase the risk in placental abruption [17].
Women that use less calcium level in diet, have more chances
of increase blood pressure in their pregnancy [18]. Deficiency
of calcium level creates more chances of preeclampsia in the
women in pregnancy. They need calcium supplementation to
decrease hypertension disorders. Effect of rotation of the head
direction of the foetus inside the uterus has a great impact on
the mode of delivery of a baby [19]. The researchers studied
the effects of rotation of the foetal head on the probable
outcome of the delivery mode.
The effects of medical complications that cause C-Section
are very important to handle with great care. In a study [20],
they discussed the scope of elective caesarean section without
any medical complications. They discussed that the rate of C-
Section deliveries is increasing day by day mostly due to the
elective mode of delivery. A survey was conducted by
interviewing the expected mothers about their own choice of
the mode of birth and most of the women selected C-Section,
because majority of women think that it is a safe way for the
baby as compared to the vaginal birth.
The maternal age of the expecting mother also influences
the mode of delivery either it will be a C-Section or a normal
birth. They performed an analysis about the impact of the
maternal age on the mode of delivery of an expecting woman
[21]. They discussed that the women with age more than 35
years have a caesarean section of about 46.1% and that of the
age between 30 to 34 years was 40.9%. They showed that only
(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 12, 2019
122 | P a g e
www.ijacsa.thesai.org
age is not the influencing factor on the mode of delivery.
Besides age there are many other factors such as stress, fatigue,
posture of working, economic status is also affecting the birth
mode of a becoming mother.
The effect of maternal age and the foetal sex are also
considered a cause of C-section in most of the expecting
women. They discussed that the foetal sex and maternal age
combine affect the mode of delivery of a woman [22]. They
showed that the risk of operative deliveries increased with a
woman of age >40 and carrying a male foetus. The risk of
maternal diabetes and increased hospitalization of the women
having age<40, but carrying a male foetus also become a
cause of operative or caesarean section delivery.
A smart decision support system is available for
performing statistical analysis for finding different results. An
analysis applied machine learning techniques on smart
decision support systems for prediction of the correct
treatment methods for a pregnant woman [23]. A comparison
between two Bayesian classifiers is performed to classify
hypertensive disorder severity among the pregnant women and
the results showed that the Bayesian classifier produces the
best results with a precision of 0.400 as compared to AODE
which has a precision of 0.275.
III. METHODOLOGY
The attributes of the dataset are listed with their types and
full description in the Table I. Besides all these attributes, a
class attribute is used for performing the classification of the
data based on the values given to the attributes about a
specific entity. Based on the values of the attributes, it is
decided that a specific entity is assigned to a class with label 0
or 1. The accuracies of the applied techniques are calculated
and the Kappa values for these accuracies are also observed.
Based on these calculations, the best methods for solving such
type of problems are suggested.
The techniques of Artificial Neural Networks (ANNs),
Support Vector Machines (SVM), Random Forests (RF),
Recursive Partitioning (RPART) and Conditional inference
Tree (CTree) are applied on the dataset. After applying these
techniques, the results are recorded and presented in the form
of tables, curves and graphs.
TABLE. I. ATTRIBUTE DESCRIPTION FOR THE DATASET
S#
Name
Description
Type
1
Age
Age of the patient
Integer (17-40)
2
Fp
First pregnancy or not
Binary
3
Pcp
No. of pregnancies before the current pregnancy
Integer
4
Lb
No. of live births
Integer
5
Boys
No. of boys
Integer
6
Girls
No. of girls
Integer
7
Abortin
No. of abortions
Integer
8
Miscag
No. of miscarriages
Integer
9
Last_mode
Mode of last delivery
Binary
10
Inherited
Husband having any inherited disease or not
Binary
11
Menstrual
Having menstrual regular or not
Binary
12
Days_mentcal
No. of days of menstruation cycle
Integer
13
Last_time
Last time of menstruation
Integer
14
Bleeding
Having bleeding or not
Binary
15
Fatigue
Feeling fatigue during pregnancy or not
Binary
16
Diabetic
Having any diabetes disease or not
Binary
17
Breathing
Having breathing issue or not
Binary
18
Headache
Having headache or not
Binary
19
Fast_bet
Having fast heart beat or not
Binary
20
Surgery
Having any surgery before or not
Binary
21
Hemoglobin
Hemoglobin level
Integer
22
BMAX
Blood pressure maximum
Integer
23
BMIN
Blood pressure minimum
Integer
24
BPORNOT
Having BP issue or not
Binary
25
Medcin
Taking any medication or not
Binary
26
Headic
Any sort of tension or not
Binary
27
Hyperten
Having any hypertension disorder or not
Binary
28
FA
Taking Folic Acid tablets or not
Binary
29
Iron
Taking iron supplements or not
Binary
(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 12, 2019
123 | P a g e
www.ijacsa.thesai.org
IV. PERFORMANCE EVALUATION
In this section, we discuss the experimental setup and tools
used for evaluation. Further, the results are presented and
discussed in detail.
A. Experimental Setup
The evaluation of algorithms is carried out by using
RStudio tool, which is the most user-friendly tool available for
performing machine learning tasks on any given dataset.
Before processing the data using RStudio, the dataset was
converted into CSV format. The evaluation results of ANN,
SVM, RF, RPART and CTREE algorithms are gained and a
comparison between the results is performed to judge the most
appropriate technique for performing such type of
classification jobs.
Further, the evaluation of the algorithms is carried out by
using the cross-validation method, which is the most
frequently used method for performing the validation on a
collected set of data for statistical analysis. The 10-fold cross
validation is the most important type of validation that is
mostly used in evaluation of different machine learning
techniques. In this method, the whole dataset is divided into
10 equal parts/folds. The evaluation is performed by selecting
one of the folds as a test set and the remaining 9 folds as the
training data. The same procedure is repeated each time for
every fold of the data set. In this way, 10 iterations will be
performed on the supplied data for validation of the results of
the accuracies required by the given machine learning techniques.
The dataset is divided into two halves of 75% and 25%
size. One is used as training dataset and the other is used for
testing the results based on the training dataset values. After
doing this process, the most accurate results about the analysis
of the data are obtained.
B. Implementation Tools
RStudio is a free open source Integrated Development
Environment tool available online for performing machine
learning tasks with the help of statistical analysis. It provides
an interactive environment that provides the facility of coding,
debugging, plotting different types of graphs and viewing the
history of the previous code run. It contains a set of built in
libraries and functions for performing different tasks. It
provides an interactive environment to the users for making
the implementation of different machine learning techniques
easy and efficient. We can perform classification and
clustering of data using RStudio. The ROC curves can be
created using RStudio for viewing the result of the applied
techniques graphically for comparison or analysis. RStudio
provides an environment to the user for performing the data
manipulation easily and efficiently. All the coding is done in
the R language by using different packages and libraries.
There are many built-in libraries and packages available with
the RStudio application software. R programming Language is
the base of the RStudio application. Any computer running
RStudio must have installed R language prior to RStudio.
C. Results
The accuracies and Kappa statistic values for the data are
calculated and the results are shown in the tabular form for
interpretation. Next the Receiver Operating Curves are created
for each technique to check the overall accuracies of the
mentioned techniques. After that graphs are plotted for
interpretation of accuracies and Kappa values of all the folds
in 10-fold cross validation. Table II shows the accuracy values
of RF, SVM, RPART, CTREE and NNET. Results show that
the accuracy of RF Algorithm is highest among all the five
mentioned techniques followed by NNET, RPART, CTREE
and SVM in the descending order. RF has highest accuracy
value of 0.9972. The accuracy of NNET is 0.9863, RPART is
0.9838, CTREE is 0.9835, and SVM is 0.9784.
Table III shows the Kappa statistics values for all the five
classifiers and provides an insight about the performance of
the classification algorithms. Results show the Kappa statistics
values for RF, SVM, RPART, CTTREE and NNET. These
values depict that Random Forest algorithm shows the highest
Kappa value of 0.9941 for the given data. The Neural Net has
second highest value 0.9710, after that the value of Recursive
Partitioning is 0.9666, then CTREE has 0.9657 and at last
SVM has the value 0.9551. These statistics show that RF
shows highest associativity among the values of different
attributes of the given dataset.
TABLE. II. ACCURACY VALUES OF CLASSIFIERS
Classifier
Min.
1stQuadrant
Median
Mean
3rd Quadrant
Max.
NA’s
RF
0.9722
1.0000
1.0000
0.9972
1
1
0
SVM
0.8947
0.9722
0.9861
0.9784
1
1
0
RPART
0.9211
0.9722
1.0000
0.9838
1
1
0
CTREE
0.9444
0.9722
0.9868
0.9835
1
1
0
NNET
0.9722
0.9724
0.9865
0.9863
1
1
0
TABLE. III. KAPPA STATISTICS VALUES OF CLASSIFIERS
Classifier
Min.
1st Quadrant
Median
Mean
3rd Quadrant
Max
NA’s
RF
0.9408
1.0000
1.0000
0.9941
1
1
0
SVM
0.7847
0.9412
0.9712
0.9551
1
1
0
RPART
0.8403
0.9423
1.0000
0.9666
1
1
0
CTREE
0.8861
0.9423
0.9721
0.9657
1
1
0
NNET
0.9417
0.9423
0.9712
0.9710
1
1
0
(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 12, 2019
124 | P a g e
www.ijacsa.thesai.org
The results of training and validation accuracies are
combined for performing a comparative analysis of training
and validation results of the data. Fig. 1 shows the comparison
of training and validation accuracies of the classification
techniques. CTREE has highest set of values for both training
and validation results of the data and RPART has highest
accuracy values. NNET with third highest accuracy, RF has
values 96.15616 and 100 for training and validation
respectively. SVM shows lowest values among all. This plot
shows that all the applied techniques have better accuracy
values for the validation data set as compared to the training
part of the data. RPART and CTREE have highest points in
this plot.
Next the ROC curves for all the five classifiers are
obtained by using the RStudio. Fig. 2 shows the overall
accuracies of NNET, SVM, RF, RPART and CTREE as a
whole. Results show the classification of the given dataset into
normal and C-Section values according to given class labels
assigned. The area under the curve shows the sensitivity over
specificity for NNET is 1.000, SVM is 0.993, RF is 1.000,
RPART is 1.000 and CTREE is 1.000. Results depict that
NNET, RPART, CTREE and RF have maximum sensitivity to
specificity ratio.
Fig. 3 depicts the training and testing accuracies of all the
techniques. Results show that in the testing phase, Tree-based
techniques are providing better accuracy values as compared
to other techniques on the given dataset.
Fig. 1. Training vs Validation Accuracies of Classifiers.
Fig. 2. Area under the Curve of All the Classifiers.
Fig. 3. Comparison of Training and Testing Accuracies of Classifiers.
Table IV shows the accuracies of the classifiers for all the
10 folds individually. Every fold performs a complete run on
the data for every technique separately and the result is
returned at the end of complete run of the data. Every fold has
specific values for the classification of the data. The value 1
shows the maximum accuracy of the classifiers.
Table V shows the Kappa values of the classifiers for all
the 10 folds in a 10-fold cross validation process. Results
show that in each iteration, as the values of training and
testing data are changed, so the results for the evaluation of
algorithms are also changed accordingly.
TABLE. IV. ACCURACIES OF CLASSIFIERS FOR 10-FOLD CROSS
VALIDATION
Folds
RF
SVM
RPART
CTREE
NNET
Fold01
1
1
1
1
1
Fold02
1
0.9722
0.9722
1
0.9722
Fold03
0.9722
0.8947
0.9210
0.9444
1
Fold04
1
1
1
0.9722
0.9729
Fold05
1
0.9722
0.9722
0.9722
1
Fold06
1
1
1
0.9722
1
Fold07
1
1
1
1
0.9722
Fold08
1
0.9722
1
1
1
Fold09
1
0.9722
0.9722
1
0.9729
Fold10
1
1
1
0.9736
0.9722
TABLE. V. KAPPA STATISTICS OF CLASSIFIERS FOR 10-FOLD CROSS
VALIDATION
Folds
RF
SVM
RPART
CTREE
NNET
Fold01
1
1
1
1
1
Fold02
1
0.9423
0.9423
1
0.9423
Fold03
0.9407
0.7847
0.8403
0.8860
1
Fold04
1
1
1
0.9423
0.9417
Fold05
1
0.9423
0.9423
0.9423
1
Fold06
1
1
1
0.9423
1
Fold07
1
1
1
1
0.9423
Fold08
1
0.9407
1
1
1
Fold09
1
0.9407
0.9407
1
0.9417
Fold10
1
1
1
0.9442
0.9423
(IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 10, No. 12, 2019
125 | P a g e
www.ijacsa.thesai.org
Fig. 4. Comparison of Accuracies and Kappa Values of Classifiers at
Confidence Level 0.95.
Fig. 4 shows the comparison of Accuracies and Kappa
Statistics values for all the five mentioned techniques. Results
show the performance evaluation of Random Forest, SVM,
NNet, RPART and CTREE for the classification accuracy of
data of the pregnant women. This comparison shows that RF
has the highest accuracy value as well as kappa value among
all. The accuracy of RF is 0.9972. the Kappa value of RF is
also highest as 0.9941. RPART and CTREE have highest
training accuracies. Except SVM, all the techniques have
maximum validation accuracies and AUC values.
V. CONCLUSION
In this paper, we show the accuracies of classification
algorithms on the dataset of pregnant women for classifying
the data into two groups based on the mode of delivery of a
woman. The techniques of Support Vector Machines, Neural
Networks, Random Forest, Recursive Partitioning and
Conditional Inference Tree are applied on the given dataset
and the results in the form of accuracy tables, ROC curves and
different plots are recorded for further interpretation of the
data.
The results show that the tree-based techniques are best
suited for the classification of data into normal and C-Section
classes as compared to the kernel based approach. Among the
tree based techniques, Random Forest shows the maximum
accuracy value for the classification of the given data. The
Accuracy of Random Forest is highest as calculated 0.9972.
The Kappa Statistics for Random Forest is also higher than all
the other implemented techniques and is calculated to be
0.9941.
It can be concluded from all the above discussion and the
presented results that the technique of Random Forest is best
suited for this type of data as it provides the maximum value
for accuracy on the given dataset.
REFERENCES
[1] Snyder, R. M. (2015). An Introduction to Topic Modeling as an
Unsupervised Machine Learning Way to Organize Text Information.
Association Supporting Computer Users in Education.
[2] Jordan, M. I., & Mitchell, T. M. (2015). Machine learning: Trends,
perspectives, and prospects. Science, 349(6245), 255-260.
[3] Holzinger, A. (2016). Interactive machine learning for health
informatics: when do we need the human-in-the-loop? Brain Informatics,
3(2), 119-131.
[4] Samuel, A. L. (1959). Some studies in machine learning using the game
of checkers. IBM Journal of research and development, 3(3), 210-229.
[5] Kotsiantis, S. B., Zaharakis, I., & Pintelas, P. (2007). Supervised
machine learning: A review of classification techniques. Emerging
artificial intelligence applications in computer engineering, 160, 3-24.
[6] Caruana, R., & Niculescu-Mizil, A. (2006). An empirical comparison of
supervised learning algorithms. Paper presented at the Proceedings of
the 23rd international conference on Machine learning.
[7] Phyu, T. N. (2009). Survey of classification techniques in data mining.
Paper presented at the Proceedings of the International MultiConference
of Engineers and Computer Scientists
[8] Krishnaiah, V., Narsimha, G., & Chandra, N. S. (2014). Survey of
classification techniques in data mining. International Journal of
Computer Sciences and Engineering, 2(9), 65-74.
[9] Dulitzki, M., Soriano, D., Schiff, E., Chetrit, A., Mashiach, S., &
Seidman, D. S. (1998). Effect of very advanced maternal age on
pregnancy outcome and rate of cesarean delivery. Obstetrics &
Gynecology, 92(6), 935-939.
[10] Sana, A., Razzaq, S., & Ferzund, J. (2012). Automated Diagnosis and
Cause Analysis of Cesarean Section Using Machine Learning
Techniques. International Journal of Machine Learning and Computing,
2(5), 677.
[11] Hueston, W. J. (1995). Site-to-site variation in the factors affecting
cesarean section rates. Archives of Family Medicine, 4(4), 346.
[12] Adashek, J. A., Peaceman, A. M., Lopez-Zeno, J. A., Minogue, J. P., &
Socol, M. L. (1993). Factors contributing to the increased cesarean birth
rate in older parturient women. American journal of obstetrics and
gynecology, 169(4), 936-940.
[13] Nazir, S. (2015). Determinants of cesarean deliveries in Pakistan.
Pakistan Institute of Development Economics.
[14] Leone, T. (2014). Demand and supply factors affecting the rising
overmedicalization of birth in India. International Journal of
Gynecology & Obstetrics, 127(2), 157-162.
[15] Malone, F. D., Canick, J. A., Ball, R. H., Nyberg, D. A., Comstock, C.
H., Bukowski, R., . . . Craigo, S. D. (2006). First-Trimester or Second-
Trimester Screening, or Both, for Down’s Syndrome. Obstetrical &
gynecological survey, 61(4), 226-228.
[16] Ludvigsson, J. F., & Ludvigsson, J. (2001). Coeliac disease in the father
affects the newborn. Gut, 49(2), 169-175.
[17] Ananth, C. V., Smulian, J. C., & Vintzileos, A. M. (1999). Incidence of
placental abruption in relation to cigarette smoking and hypertensive
disorders during pregnancy: a meta-analysis of observational studies 1.
Obstetrics & Gynecology, 93(4), 622-628.
[18] Sultana, R., Singh, K. R., & Joshi, V. Original Research Article Role of
Serum Calcium Level in Pregnancy Induced Hypertension.
[19] Romano, A. M. (2007). Research summaries for normal birth. The
Journal of perinatal education, 16(3), 53.
[20] Miesnik, S. R., & Reale, B. J. (2007). A review of issues surrounding
medically elective cesarean delivery. Journal of Obstetric, Gynecologic,
& Neonatal Nursing, 36(6), 605-615.
[21] Benli, A. R., Benli, N. C., Usta, A. T., Atakul, T., & Koroglu, M. (2015).
Effect of maternal age on pregnancy outcome and cesarean delivery rate.
Journal of clinical medicine research, 7(2), 97.
[22] Weissmann-Brenner, A., Simchen, M. J., Zilberberg, E., Kalter, A., &
Dulitzky, M. (2015). Combined effect of fetal sex and advanced
maternal age on pregnancy outcomes. Medical science monitor:
international medical journal of experimental and clinical research, 21,
1124.
[23] Moreira, M. W., Rodrigues, J. J., Oliveira, A. M., Saleem, K., & Neto, A.
V. (2016). An inference mechanism using bayes-based classifiers in
pregnancy care. Paper presented at the e-Health Networking,
Applications and Services (Healthcom), 2016 IEEE 18th International
Conference on.