Child Mortality Prediction using Machine Learning

Abstract
 Children’s Mortality alludes to mortality of children younger than 5. The kid death rate, in addition
under-five death rate, alludes to the probability of biting the mud among birth and exactly 5 years
recent.

 The purpose is to analysis AI based mostly strategies for grouping of mortality vertebrate upbeat
characterization brings concerning best truth.

 The examination of dataset by directed AI procedure (UN) to catch a couple of data\'s like,
variable characteristic proof, uni-variate investigation, bi-variate and multi-variate examination,
missing value medicines and dissect the data approval, data cleaning/getting prepared and
knowledge illustration are done on the entire given dataset.

 To propose AN AI based mostly and moreover, to seem at and examine the presentation of various
AI calculations for the given dataset.
Existing System

 supervised learning method you would come up with if you wanted to model a
predictive modeling problem probabilistically. Naive bayes simplifies probability
calculations by assuming that the probability.

 each attribute belongs to a given class value is independent of all other attributes.
This is a strong assumption but leads to a quick and efficient method. The
probability of a class value giving the value of an attribute is called the
conditional probability. By multiplying the conditional probabilities for each
attribute for a given class value, we get the probability that a data instance
belongs to that class. To make a prediction,
Drawback in Existing System

 May need more memory space.

 Requires more effort during design.

 May require more run time

Proposed System
 The proposed model is to build a model to predict mortality.

 Collected data may contain missing values.

 To get better results, the data should be preprocessed to improve the efficiency of the
algorithm.

 The data set collected to predict the given data is divided into training set and test set. In
general, a ratio of 7:3 is applied to divide the training set and the test set.

 The data model created using machine learning algorithms is applied to the training set,
and based on the accuracy of the test results, the prediction of the test set is made.

 The model can classify mortality. Different machine learning algorithms can be compared
and the best algorithm can be used for classification.
Advantages

 Consistency. The fact that the engine allows all applications to run using the same
base RAD applications brings consistency to the Service Manager application
suite

 Reduced Development Times

 Flexibility
System Requirement
Hardware Specification

 Processor : i3 Processor

 Hard disk : 250 GB

 RAM : 4 GB

Software Specification

 Operating System : windows 10

 Frontend : Java Mavan with Python

 IDE Tool : Pycharm

 Back End : Mysql

System architecture
Accepting all Information
Web Server
Service Provider
Login,
D
Train & Test child care Content Data Sets
Accessing Dataatasets Results Storage
View Trained and Tested Accuracy in Bar Chart
Process all user queries View Trained and Tested Accuracy Results
View Prediction Of child care Content Data Sets
View Child care Content Data Sets Ratio
Store and retrievals
Download Predicted Data Sets
View child Content Data Sets Ratio Results
Data Base
Server View All Remote Users.

Register and login,

Remote User
Predict Child care content type,
View your profile.
Structure Chart
Train & Test Child care Content Data
Sets, View Trained and Tested
Accuracy in Bar Chart View Trained
and Tested Accuracy Results

Service Provider Login

System predict Child care
content type

View Prediction of Child care

Content Data Sets, View Drug Register and
Content Data Sets Ratio, Login with the
Download Predicted Data Sets system View you’re profile

Remote User
View Child care Content Data Sets
Ratio Results, View All Remote
Users.
Data Flow Diagram
Level 1
New
Check Patient
User New TblPatient
Data Prediction Data
Required
Details Submit

Extract
Dastaset
Level 2

Service Provider

Account Login

Patient Dataset

Child care Information

Data Collection

Prediction
 Dataset
collection
Identify TblPatient
Patient and
Patient Data Collection Entry Partition
Methodologies Description
 Data Collection

 Data Preprocessing

 Mixed test

 Blind test

 Prediction of unknown Child care line response

 Investigation of genomic aberration contribution to the response

Data Collection
Service provider

in this module, the service provider has to login by using valid user name and password. after login
successful he can do some operations such as login, train & test drug content data sets, view trained and
tested accuracy in bar chart

view and authorize users

in this module, the admin can view the list of users who all registered. in this, the admin can view the
user’s details such as, user name, email, address and admin authorizes the users.

remote user

in this module, there are n numbers of users are present. user should register before doing any
operations. once user registers, their details will be stored to the database. after registration successful,
he has to login by using authorized user name and password. once login is successful user will do some
operations like register and login, predict drug content type, and view your profile.
Data Pre-Processing

 Machine learning validation techniques are used to obtain the error rate of a machine learning (ML)
model, which can be considered close to the actual error rate of the data set. If the data volume is large
enough to represent the set, you may not need validation techniques. However, in real-world situations,
working with data samples may not be a true representation of a given data set. To find the missing
value, double the value and description of the data type, whether it's a float variable or an integer.
Mixed test

This experiment evaluates the performance of models in known Child care

line pairs. Of all 211,404 possible Child care line pairs, GDSC provides the response
for 172,114 pairs [33]. The data is shuffled before splitting to help the model remains
general and reduces over fitting. The known pairs are split into 80% as the training
set, 10% as the validation set and 10% as the testing set. While the validation set is
used to modify the hyper parameter of the model in the training phase, the testing set
is used to evaluate the performance of the model.
Blind Test

a drug appeared in the testing set may also appear in the training phase.
However, we sometimes need to predict the response of a new drug, for example,
newly invented one. to evaluate the prediction performance of unseen drugs. Drugs
are constrained from existing in training and testing at the same time. Of 90%
(201/223) drugs, their IC50 values are randomly selected for training, including 80%
drugs for the training set and 10% drugs for the validation set
Prediction of unknown Child care line
response

predicting missing Child care line response. The best pre-trained model in the mixed
test experiment is used to predict missing pairs in UNICEF dataset. Then we select
top 10 drugs that have the lowest and highest IC50 values to further investigate
whether drugs having lower IC50 are more effective to the treatment of cancer,
meanwhile the ones having higher IC50 are less effective.
Investigation of genomic aberration
contribution to the response
Given a pair of drug and cell-line, the saliency value is calculated by taking
derivative of predicted response with respect to the cell-line, in one-hot vector
format. To do this experiment, we chose a drug, which has the lowest average IC50
among all drugs in the prediction of unknown childcare line response. Then selected
three cell-lines that have lowest IC50 values to that drug. Then, for each pair, we
calculate the saliency value then take ten most important genomic aberrations. After
that, we find evidence to support the contribution of the genomic aberrations to the
response of that drug on the selected cell-lines.