Here in this ppt we are showing how we are using machine learning to predict cardio d functionalities or diseases in a patient before he is diagnosed with the disease or in the later stages.
3. PROBLEM STATEMENT
INTRODUCTION
Heart diseases are the primary reason of deaths in adults.
In the last 10 years, from 2012 to 2021, there had been
a 54 per cent increase in heart attack deaths. Last year,
70 per cent of heart attack deaths occurred in the 30-60
age group. A total of 19,744 people aged 30 to 60 years
died due to heart attacks in 2021. Thus we need to find a
Method to predict any possibilities of CVDs.
NEXT
4. OBJECTIVE
Objective of this project is to check whether the patient
is likely to be diagnosed with any CVDs based on their
age, sex, chest pain, blood pressure, cholesterol.... Etc,
Using a classification model and enhancing its frequency.
INDEX
5. BACKGROUND
DATASET
S.No Attribute Name Description Range of values
1. Age Age of the person in years 28 to 77
2. Sex Gender of the person M, F
3. Chest Pain Type Typical Angina ,Atypical
Angina
,Non-angina pain,
Asymptomatic
4. RestingBP Blood Pressure in mm Hg 80 to 200
5. Cholestrol Serum cholesterol in mg/dl 0 to 603
6. FastingBS Fasting Blood Sugar in
mg/dl(1: >1200: < 120)
0, 1
7. RestingECG Electrocardiographic Results N, ST, LVH
8. Max HR Maximum Heart Rate Achieved 60 to 202
9 Exercise angina Exercise Induced Angina Y, N
10. Oldpeak ST depression induced by
exercise relative to res
-2.6 to 6.2
11. ST_Slope Slope of the Peak Exercise ST
segment
Up, Flat, Down
NEXT
6. RELATED WORK
• Robert Detrano, Andras Janosi, Walter Steinbrunn, Matthias Pfisterer, Johann-Jakob Schmid,
Sarbjit Sandhu, Kern H. Guppy, Stella Lee and Victor Froelicher in 1989 used the logistic
regression algorithm and predicted the dataset with 77% accuracy.
• In 2001 used multiple algorithms to predict the heart disease and came up with the accuracy of
81.48% using the Naïve Bayes algorithm.
• Polat, K., Sahan, S., Kodaz, H., Günes, S in 2005 used the AIS algorithm and came up with the
accuracy of 84.5%.
• Polat, K., S. Sahan, and S. Gunes in 2007 used Fuzzy-AIRS-KNN 8 algorithm and increased
het accuracy of model BY 87.0% .
• Lamia Abed Noor Muhammed in 2012 used the CLIP4 ensemble to increase the accuracy of
the model by 90.4% .
INDEX
7. Feature Extraction
Patient data and
health records
Medical Dataset
Machine Learning
Techniques
Linear Regression
Random Tree
Naïve Bayes
Discriminant Analysis
Ensemble
Performance Evaluation
Training Dataset
Testing Dataset
Selecting Best Model
Cardiovascular Prediction
FRAMEWORK
NEXT
8. RANDOM FOREST CLASSIFIER
● Random forests or random decision forests is a learning
method for classification.
● Like its name implies, it consists of a large number of
individual decision trees that operate as an group.
● Each individual tree in the random forest spits out a
class prediction and the class with the most votes
becomes our model’s prediction
INDEX
9. RESULT
MODEL ACCURACY
rf Random Forest Classifier 91.20 %
gbc Gradiant Boosting Classifier 89.97%
lightgbm Light Gradiant Boosting
Classifier
89.56%
Following pie chart shows that from the given data there were 42.14%
people diagnosed negative and 57.86% people diagnosed positive
with heart disease.Other pie chart shows, from the given data there
were 42.43% people predicted negative and 58.57% people predicted
positive with heart disease.
NEXT
10. • It is found that the group between 50 and 65 had the highest frequency/occurance
in the given dataset.
• It was found that out of total positive cases of heart disease, 77% of the cases
showed no symptom of any chest pain i.e., they were asymptomatic.
INDEX
11. FUTURE SCOPE
It should be recognized that, although prevention will delay or even
prevent the
onset of CVD and the cost of cardiovascular treatment, patients will
need
medical care longer and the lifetime cost of care may not be reduced
as patients
live longer. Thus, prevention strategies should not be evaluated
solely on their
ability to reduce cost of care, but instead they should be valued
based on a
combination of cost and impact on patient well being, including the
length and
quality of life.
NEXT